University of Central Florida STARS
Electronic Theses and Dissertations, 2004-2019
2008
Laser Enhanced Doping For Silicon Carbide White Light Emitting Diodes
Sachin Bet University of Central Florida
Part of the Materials Science and Engineering Commons Find similar works at: https://stars.library.ucf.edu/etd University of Central Florida Libraries http://library.ucf.edu
This Doctoral Dissertation (Open Access) is brought to you for free and open access by STARS. It has been accepted for inclusion in Electronic Theses and Dissertations, 2004-2019 by an authorized administrator of STARS. For more information, please contact [email protected].
STARS Citation Bet, Sachin, "Laser Enhanced Doping For Silicon Carbide White Light Emitting Diodes" (2008). Electronic Theses and Dissertations, 2004-2019. 3680. https://stars.library.ucf.edu/etd/3680
LASER ENHANCED DOPING FOR SILICON CARBIDE WHITE LIGHT EMITTING DIODES
by
Sachin Madhukar Bet B.E. Govt. College of Engineering. (COEP), India, 2000 M.S. University of Central Florida, USA, 2003
A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Mechanical, Materials and Aerospace Engineering in the College of Engineering and Computer Science at the University of Central Florida, Orlando, Florida
Fall Term 2008
Major Professor: Aravinda Kar
1
© 2008 Sachin Madhukar Bet
2
ABSTRACT
This work establishes a solid foundation for the use of indirect band gap
semiconductors for light emitting application and presents the work on development of
white light emitting diodes (LEDs) in silicon carbide (SiC). Novel laser doping has been
utilized to fabricate white light emitting diodes in 6H-SiC (n-type N) and 4H-SiC (p-type
Al) wafers. The emission of different colors to ultimately generate white light is tailored
on the basis of donor acceptor pair (DAP) recombination mechanism for luminescence.
A Q-switched Nd:YAG pulse laser (1064 nm wavelength) was used to carry out
the doping experiments. The p and n regions of the white SiC LED were fabricated by
laser doping an n-type 6H-SiC and p-type 4H-SiC wafer substrates with respective
dopants. Cr, B and Al were used as p-type dopants (acceptors) while N and Se were used
as n-type dopants (donors). Deep and shallow donor and acceptor impurity level states
formed by these dopants tailor the color properties for pure white light emission.
The electromagnetic field of lasers and non-equilibrium doping conditions enable
laser doping of SiC with increased dopant diffusivity and enhanced solid solubility. A
thermal model is utilized to determine the laser doping parameters for temperature
distribution at various depths of the wafer and a diffusion model is presented including
the effects of Fick’s diffusion, laser electromagnetic field and thermal stresses due to
localized laser heating on the mass flux of dopant atoms. The dopant diffusivity is
calculated as a function of temperature at different depths of the wafer based on measured
dopant concentration profile. The maximum diffusivities achieved in this study are
4.61×10-10 cm2/s at 2898 K and 6.92×10-12 cm2/s at 3046 K for Cr in 6H-SiC and 4H-SiC respectively. Secondary ion mass spectrometric (SIMS) analysis showed the
3 concentration profile of Cr in SiC having a penetration depth ranging from 80 nm in p-
type 4H-SiC to 1.5 μm in n-type 6H-SiC substrates respectively. The SIMS data
revealed enhanced solid solubility (2.29×1019 cm-3 in 6H-SiC and 1.42×1919 cm-3 in 4H-
SiC) beyond the equilibrium limit (3×1017 cm-3 in 6H-SiC above 2500 °C) for Cr in SiC.
It also revealed similar effects for Al and N. The roughness, surface chemistry and
crystalline integrity of the doped sample were examined by optical interferometer, energy
dispersive X-ray spectrometry (EDS) and transmission electron microscopy (TEM)
respectively. Inspite of the larger atomic size of Cr compared to Si and C, the non-
equilibrium conditions during laser doping allow effective incorporation of dopant atoms
into the SiC lattice without causing any damage to the surface or crystal lattice. Deep
Level Transient Spectroscopy (DLTS) confirmed the deep level acceptor state of Cr with
activation energies of Ev+0.80 eV in 4H-SiC and Ev+0.45 eV in 6H-SiC. The Hall Effect
measurements showed the hole concentration to be 1.98×1019 cm-3 which is almost twice
the average Cr concentration (1×1019 cm-3) obtained from the SIMS data. These data
confirmed that almost all of the Cr atoms were completely activated to the double
acceptor state by the laser doping process without requiring any subsequent annealing step.
Electroluminescence studies showed blue (460-498 nm), blue-green (500-520 nm) green (521-575 nm), and orange (650-690 nm) wavelengths due to radiative recombination transitions between donor-acceptors pairs of N-Al, N-B, N-Cr and Cr-Al respectively, while a prominent violet (408 nm) wavelength was observed due to transitions from the nitrogen level to the valence band level. The red (698-738 nm)
luminescence was mainly due to metastable mid-bandgap states, however under high
4 injection current it was due to the quantum mechanical phenomenon pertaining to band broadening and overlapping. This RGB combination produced a broadband white light spectrum extending from 380 to 900 nm. The color space tri-stimulus values for 4H-SiC doped with Cr and N were X = 0.3322, Y = 0.3320 and Z = 0.3358 as per 1931 CIE
(International Commission on Illumination) corresponding to a color rendering index of
96.56 and the color temperature of 5510 K. And for 6H-SiC n-type doped with Cr and Al, the color space tri-stimulus values are X = 0.3322, Y = 0.3320 and Z = 0.3358. The CCT was 5338 K, which is very close to the incandescent lamp (or black body) and lies between bright midday sun (5200 K) and average daylight (5500 K) while CRI was
98.32. Similar white LED’s were also fabricated using Cr, Al, Se as one set of dopants and B, Al, N as another.
5 ACKNOWLEDGMENTS
I would like to express my deepest gratitude towards Dr. Aravinda Kar, the best
advisor and teacher I could have wished for. He is actively involved in the work of all his
students, and clearly always has their best interest in mind. It was a pleasure working
under his supervision and I appreciate his constant guidance, encouragement and support.
His curiosity of why followed by how not only allowed me to manage difficulties and accomplish challenging research projects but also provided an appetite to think independently and creatively. I would like to specially thank Dr. Nathaniel Quick, president of AppliCote Associates, LLC. for his constant support, guidance and encouragement in achieving my research goals. I would like to thank all the other committee members Dr. Neelkanth Dhere, Dr. Peter Delfyett, Dr. Winston Schoenfeld and Dr. Raj Vaidyanathan for serving on my final examination committee.
A special word of thanks to AppiCote Associates, LLC., Florida Photonics Centre of Excellence (FPCE) and CREOL/UCF for supporting this research work and II-VI
Incorporated for supplying SiC wafers. Thanks to all the technical staff at Advanced
Materials Characterization Facility (AMPAC-MCF) for their help in the materials
characterization work and especially Michael Klimov for assistance with SIMS. I am
grateful to Dr. Woo Kyoung Kim and Dr. Tim Anderson at chemical engineering
department of university of Florida, Gainesville for DLTS measurements. I am also thankful to Ghanshyam Londhe and Hani Khallaf for metal contacts deposition and Arun
Vijaykumar for optical microscopy measurements. Sincere thanks to Dr. Pieter Kik,
Forrest Ruhge and Jeremy Mares (Nanophotonics Device group at CREOL) for
assistance with EL measurements. The assistance of Alan Tripak at Optronic
6 Laboratories, Inc. in Orlando is highly appreciated for the spectral power output and color index measurement of LEDs. I would like to thank all my colleagues at LAMMMP
(Chong Zhang, Danyong Zeng, Zhaoxu Tian, and Geunsik Lim) for all their support and encouragement throughout my work.
I enjoyed working at LAMMP CREOL at University of Central Florida. Lastly, and most importantly; I wish to thank my family and all my friends at UCF, FSEC,
COEP and well wishers for their constant love, encouragement and support.
7
Dedicated to my family and my late grandparents
Tukaram Venkatesh Bet (Tata)
Sulochana Tukaram Bet (Avva)
8 TABLE OF CONTENTS
LIST OF FIGURES ...... 14
LIST OF TABLES ...... 23
LIST OF SYMBOLS/ABBREVIATIONS ...... 26
TECHNICAL PATENTS, PAPERS AND PRESENTATIONS ...... 30
CHAPTER 1: INTRODUCTION ...... 35
1.1 History of LED Development ...... 35
1.2 Wide bandgap (WBG) Semiconductors ...... 38
1.2.1 Silicon Carbide ...... 39
1.2.2 Silicon Carbide LED Development ...... 42
1.3 Motivation for white LED ...... 44
1.4 Human Eye Response ...... 46
1.5 Color Rendering Index (CRI) ...... 47
1.6 Correlated Color Temperature (CCT) ...... 48
1.7 Current white LED technology based on direct bandgap semiconductors ...... 49
1.7.1 Blue LED and yellow phosphor ...... 49
1.7.2 Red plus green plus blue LEDs ...... 49
1.7.3 Red, green, and blue quantum dots in a single LED ...... 50
1.7.4 Near-UV or blue LED plus red, green, and blue phosphors ...... 50
1.8 Indirect bandgap semiconductor (SiC) for white LED ...... 51
1.9 Objectives ...... 53
1.10 Technical approach to achieve the objectives ...... 54
CHAPTER 2: LASER DOPING OF SILICON CARBIDE ...... 55
9 2.1 Laser interaction with silicon carbide ...... 55
2.1.1 Material (SiC) ...... 56
2.1.1.1 Physical properties of SiC substrates ...... 56
2.1.1.2 Thermophysical and optical properties ...... 57
2.1.2 Dopant precursors ...... 58
2.1.3 Laser characteristics for doping ...... 58
2.1.4 Doping Methods ...... 60
2.1.4.1 Conventional doping (ion-implantation, thermal diffusion) ...... 60
2.1.4.2 Laser doping...... 61
2.2 Thermal model for selection of laser doping parameters ...... 62
2.3 Laser Doping Experiment ...... 66
2.3.1 Sample Preparation ...... 66
2.3.2 Experimental Setup ...... 66
2.3.3 N-type doping (N, Se) ...... 67
2.3.4 P-type doping (Al, Cr and B) ...... 68
CHAPTER 3: CRYSTALLINE QUALITY, ELECTRONIC AND ELECTRICAL
PROPERTIES ANALYSIS ...... 71
3.1 Crystalline quality analysis ...... 71
3.1.1 Effect of laser doping on crystalline quality ...... 71
3.1.1.1 Surface roughness and chemistry analysis using optical profilometer and
EDS ...... 72
3.1.1.2 Crystal lattice analysis using FIB and TEM ...... 73
3.1.2 SIMS studies for dopant concentration and solid solubility analysis ...... 77
10 3.1.2.1 Enhancement of solid solubility for all the dopants ...... 77
3.1.2.2 Diffusion of large sized atom (Cr) in SiC ...... 79
3.1.2.3 Diffusion of Al, Cr and N for SiC White LEDs ...... 80
3.2 Electronic Properties Characterization ...... 83
3.2.1 Analysis of Cr energy levels and electronic defect states using DLTS ...... 83
3.2.2 Analysis of activated state of dopants using Hall Effect measurement ...... 89
3.3 Electrical Characterization ...... 90
3.3.1 P-N junction device fabrication for study of electrical properties ...... 90
3.3.2 Measurement of C-V characteristics using LCR meter ...... 91
3.3.3 Measurement of I-V Characteristics using I-V curve tracer ...... 93
3.3.3.1 Forward bias characteristics ...... 93
3.3.3.2 Reverse I-V characteristics ...... 98
CHAPTER 4: MODELLING OF DIFFUSION COEFFICIENT ...... 99
4.1 Laser enhancement of diffusion ...... 99
4.2 Theoretical model ...... 101
4.3 Experimental results ...... 103
CHAPTER 5: SIC WHITE LIGHT EMITTING DIODES ...... 110
5.1 Indirect bandgap SiC ...... 110
5.2 Mechanism for White Light Emission in SiC ...... 111
5.2.1 Single donors/acceptors (N / Al, B) ...... 112
5.2.2 Double donors/acceptors (Se / Cr) ...... 113
5.2.3 Flow of electron and hole for DAP recombination under applied bias ...... 114
5.3 White LEDs in 6H-SiC (n-type-N) wafers with Cr and Al ...... 117
11 5.4 White LEDs in 6H-SiC (n-type-N) wafers with B and Al ...... 120
5.5 White LEDs in 4H-SiC (p-type-Al) wafers with Cr and N ...... 121
5.6 White LEDs in 4H-SiC (p-type-Al) wafers with Cr and Se ...... 124
5.7 CRI and CCT for SiC White LEDs ...... 125
5.8 Power Output of SiC White LED’s ...... 127
CHAPTER 6: FACTORS AFFECTING THE PERFORMANCE OF SiC WHITE LEDS
...... 129
6.1 Device structure, contact metal type and contact configuration ...... 129
6.2 Tuning of color temperature to obtain pure white light ...... 133
6.2.1 Color temperature tuning with a green laser ...... 133
6.2.2 Color temperature tuning with two device approach ...... 136
6.3 Power scaling of white LED in the visible regime ...... 139
6.3.1 Effect of high voltage on SiC LED output in the visible range ...... 139
6.3.2 Effect of high injection current on SiC LED output in the visible range ...... 141
6.3.2.1 Quantum mechanical effect ...... 145
6.3.2.1 Classical effect ...... 145
6.3.3 Volume/Area effect for visible power output scaling ...... 146
6.3.4 Two or multiple device output for visible power output scaling ...... 148
CHAPTER 7: SUMMARY...... 151
7.1 Conclusions ...... 151
7.2 Future Work ...... 152
APPENDIX A: PROGRAM FOR X-Y SCANNING DURING LASER DOPING ...... 154
Program ...... 155
12 APPENDIX B : LASER DOPING OF GAP, SI, SIC WITH N, AL, PD AND B ...... 157
APPENDIX C: STUDY OF CONTACT ELECTRODES TO SIC...... 163
C.1. I-V characteristics ...... 164
C.2 Thermal effects during current injection ...... 166
C.3.Contact metal degradation during EL measurements under high injection ...... 168
APPENDIX D:EXPRESSION FOR ELECTROMAGNETIC FORCE OF THE LASER
BEAM ...... 171
APPENDIX E: CONCEPT OF PHONON DOPING ...... 174
APPENDIX F: PERTURBATION OF OPTICAL PROPERTIES OF SIC ...... 177
APPENDIX G: SOLAR CELL APPLICATION-SIC ...... 181
G.1 Photocurrent Measurement ...... 182
G.2 Concept for optimization of solar cell performance ...... 185
APPENDIX H: LASER THIN FILM DEPOSITION ON PLASTIC SUBSTRATES
USING SILICON NANOPARTICLES ...... 186
APPENDIX I: LASER DOPING OF GE124 QUARTZ SUBSTRATE ...... 188
APPENDIX J: LASER DOPING OF SI WAFERS FOR FABRICATION OF TUNABLE
FREQUENCY SELECTIVE SURFACES ...... 193
J.1 Introduction ...... 194
J.2 Proposed design concept ...... 194
J.3 Quantum mechanical interpretation of FSS performance ...... 197
LIST OF REFERENCES ...... 201
13 LIST OF FIGURES
Figure 1.1. History of LED Development [Press release (2006)] ...... 37
Figure 1.2. Operating temperature of p-n junctions for semiconductors as a function of
bandgap [Pearton (2000)]...... 39
Figure 1.3. Schematic representation of various recombination processes (a) free
electron-hole recombination, (b) recombination of donor, acceptors or
isoelectronic traps (c) recombination of two bound carriers and (d)
recombination within a localized center ...... 42
Figure 1.4. Human Eye’s response to light [Robinson et. al. (1984)] ...... 47
Figure 2.1 Phase diagram in Si-C the system. α is a solid solution of C in Si and β is a
solid solution of Si in C. [Tairov et. al. (1988)] ...... 56
Figure 2.2. Temperature dependence of the diffusion coefficient for various dopants in
SiC [Salama (2003)]...... 61
Figure 2.3 Transient temperature distributions along the depths of n-type 6H-SiC and p-
type 4H-SiC wafers for laser irradiation times of t = 64 and 89 ns respectively.
...... 63
Figure 2.4. Transient temperature distributions along the depths of n-type 6H-SiC and p-
type 4H-SiC wafers at times t = 70 and 91 ns respectively after the laser
irradiation began...... 65
Figure 3.1. (a) Optical interferometric micrograph of laser Cr-doped 4H-SiC (p-type)
wafer surface showing no surface damage and improved surface roughness after
laser doping. (b) TEM micrograph showing the cross-section of the sample and
14 the path for EDS line scan. (c) EDS scan starting from A in the Au-Pd thin film
region to D in the crystalline SiC region, showing the chemistry of SiC wafer. 73
Figure 3.2 (a) TEM micrograph of the laser Cr doped 4H-SiC substrate. (b) 16 nm thick
amorphous layer, Au-Pd film and SiC lattice. (c) High resolution image is SiC
lattice extending from the amorphous region in (b) into the SiC lattice and TEM
diffraction pattern indicating a single crystal SiC pattern with no signs of
amorphization or defect generation post laser doping...... 75
Figure 3.3. SIMS analysis for the concentration profiles of laser doped Cr, along the
depth of the 6H-SiC (n-type) and 4H-SiC (p-type) substrate. The penetration
depth is 1.5 μm in the case of 6H-SiC and 80 nm in the case of 4H-SiC...... 78
Figure 3.4. SIMS analysis of white LED fabrication on 6H-SiC (n-type N) substrate laser
doped with Cr and Al showing the variation of concentration along the depth of
the substrate...... 81
Figure 3.5. SIMS analysis of white LED fabrication on 4H-SiC (p-type Al) substrate laser
doped with Cr and N showing the variation of concentration along the depth of
the substrate...... 82
Figure 3.6 (a) DLTS spectrum for 6H-SiC (n-type) substrate doped with Cr. A positive
DLTS signal was observed indicating that Cr is more of an acceptor impurity
with holes (minority carriers) as traps. (b) Activation energy calculation from
the DLTS signals in (a) for different delay times showing that Cr forms an
acceptor level of Ev + 0.458 eV in 6H-SiC...... 87
Figure 3.7 (a) DLTS spectrum for 4H-SiC (p-type) substrate doped with Cr. A positive
DLTS signal was observed indicating that Cr is more of an acceptor impurity
15 with holes (minority carriers) as traps. (b) Activation energy calculation from
the DLTS signals in (a) for different delay times showing that Cr forms an
acceptor level of Ev + 0.8 eV in 4H-SiC...... 88
Figure 3.8 Laser-doped device geometry: (a) p-n diode with laser-doped p (Al) region
and as- received n region of the n-type 6H-SiC substrate and (b) p-n diode with
laser-doped p (Al) region and as-received n region, and an additional laser-
doped n (N) region for improved contact and LED performance...... 90
The experimental parameters used for doping of these samples are listed in Table 3.2 and
Figure 3.9 show the SIMS concentration profiles for these samples...... 91
Figure 3.9 SIMS analyses of p-n junction diodes and blue LED sample showing the
variation of aluminum concentration with the depth of the substrate...... 91
Figure 3.10 C-V characteristics of LED structure (Figure 3.8b) on 6H-SiC (n-type)
substrate doped with aluminum ...... 92
Figure 3.11 I-V characteristics of different device structures: (a) as-received 6H-SiC (n-
type) substrate with indium contacts, and p-n junction diode fabricated using the
6H-SiC (n-type) substrate doped with aluminum; (b) comparison of I-V curves
of three p-n junction diodes with different aluminum concentration profiles. ... 94
Figure 3.12 Detailed I-V characteristics of a blue LED structure on 6H-SiC (n-type)
substrate doped with aluminum showing the onset of radiative recombination in
(a) and a magnified view of the onset region in (b)...... 96
Figure 4.1. The mechanism of dopant diffusion due to the concentration gradient, laser
electromagnetic field (E. M. F.) and thermal stresses in SiC during laser doping.
...... 100
16 Figure 4.2 Curve fitted SIMS concentration profile for Cr along the depths of (a) n-type
6H-SiC and (b) p-type 4H-SiC wafers...... 104
Figure 4.3. (a) Variation in the diffusivity (D) of Cr at different temperatures (T)
corresponding to different depths of the n-type 6H-SiC wafer. (b) Linear fit of
the ln (D) versus 1/T plot to obtain the activation energy Q and pre-exponential
diffusivity D0 for the diffusion of Cr...... 105
Figure 4.4 (a) Variation in the diffusivity (D) of Cr at different temperatures (T)
corresponding to different depths of the p-type 4H-SiC wafer. (b) Linear fit of
the ln(D) versus 1/T plot to obtain the activation energy Q and pre-exponential
diffusivity D0 for the diffusion of Cr...... 106
Figure 4.5 Diffusive mass fluxes of Cr due to three types of driving forces: concentration
gradient, laser electromagnetic field (E. M. F.) and thermal stresses for (a) n-
type 6H-SiC wafer and (b) p-type 4H-SiC wafer...... 108
Figure 5.1. Energy levels of multiple dopants in 6H-SiC for DAP recombination
mechanism for RGB emission...... 111
Figure 5.2. Position of Fermi-level for various regions of 4H-SiC (p-type-Al) white LED
sample doped with Se and Cr...... 115
Figure 5.3. Alignment of the Fermi-levels under equilibrium conditions for p and n
regions of 4H-SiC (p-type) white LED sample doped with Cr and Se ...... 115
Figure 5.4. Flow of electron and holes across the device under low or medium forward
bias...... 116
Figure 5.5. Flow of electron and holes across the device under very high forward bias.117
17 Figure 5.6. Electroluminescence spectrum, device structure and blue light emission
observed from a blue LED fabricated on 6H-SiC (n-type-N) substrate by laser
doping Al...... 118
Figure 5.7. Electroluminescence spectrum, device structure and red-green light emission
observed from a LED fabricated on 6H-SiC (n-type-N) substrate by laser doping
Cr. Under high injection the same green LED turns red due to a metastable mid
bandgap defect states and quantum mechanical effect...... 118
Figure 5.8. Electroluminescence spectrum, device structure and white light emission
observed from a LED fabricated on 6H-SiC (n-type-N) substrate by laser doping
Al and Cr. The DAP recombination mechanism yields RGB emission which
combine to form pure white light...... 119
Figure 5.9. Electroluminescence spectrum, device structure and blue-green light emission
observed from a LED fabricated on 6H-SiC (n-type-N) substrate by laser doping
B...... 120
Figure 5.10. Electroluminescence spectrum and device structure of a White LED
fabricated on 4H-SiC (p-type-Al) substrate by laser doping B and Al...... 121
Figure 5.11. Electroluminescence spectrum, device structure and orange light emission
observed from LED fabricated on 4H-SiC (p-type-Al) substrate by laser doping
with Cr...... 122
Figure 5.12. Electroluminescence spectrum and device structure of a green LED
fabricated on 4H-SiC (n-type-N) substrate by laser doping Cr. More of red
emission is observed due to a metastable mid bandgap defect states and quantum
mechanical effect...... 122
18 Figure 5.13. Electroluminescence spectrum, device structure and white light emission
observed from a LED fabricated on 4H-SiC (p-type-Al) substrate by laser
doping Cr and N...... 123
Figure 5.14. Electroluminescence spectrum, device structure and bluish light emission
observed from LED fabricated on 4H-SiC (p-type) substrate by laser doping it
with Se...... 124
Figure 5.15. Electroluminescence spectrum, device structure and white light emission
observed from LED fabricated on 4H-SiC (p-type) substrate by laser doping it
with Cr and Se...... 125
Figure 5.16. Laser-fabricated SiC white LED showing the color space tristimulus values
as per 1931 CIE chromaticity at 2 degree on 6H-SiC (n-type-N) wafer substrate
laser doped with Al and Cr. Point A corresponds to the white light obtained
from the emitted RGB combination in comparison to the normalized reference
point while point B represents the dominant wavelength (565 nm) in the emitted
light output...... 126
Figure 5.17. Laser-fabricated SiC white LED showing the color space tristimulus values
as per 1931 CIE chromaticity at 2 degree on 4H-SiC (p-type-Al) wafer substrate
laser doped with N and Cr. Point B represents the dominant wavelength (474.9
nm) in the emitted light output...... 127
Figure 6.1. Different contact configurations across the device with Al foil, Al plate,
Silver (Ag) pins, Copper (Cu) posts and Tungsten (W) probes...... 130
Figure 6.2. Experimental setup for determining the % of green component required to
improve the color temperature from 0 K to 5500 K ...... 134
19 Figure 6.3. Graph of ratio of green laser power output to the device output vs. the color
temperature. ~0.5-1 % of green component is required to improve the color
temperature from 0 K to 5500 K ...... 135
Figure 6.4 shows the experimental setup and connection configuration for color
temperature tuning with two devices...... 136
Figure 6.5. Color temperature tuning with two device approach ...... 138
Figure 6.6. Power scaling with two device approach ...... 138
Figure 6.7. Power scaling with the high voltage in the visible range under same high
injection current ...... 140
Figure 6.8. Power output scaling linearly with the injected current in the visible range
upto 200 mA...... 142
Figure 6.9. Heating of SiC wafer substrate on induction heater upto 620°C and in a batch
furnace upto 1200°C to determine if it glows at these temperatures...... 143
Figure 6.10. Study of the observed NIR emission in SiC white LED under high injection
current ...... 144
Figure 6.11. Device and connection configurations for the study of volume/area effect for
scaling in SiC LEDs...... 147
Figure 6.12. Scaling of the power output in the visible region proportionately with
volume of the device...... 148
Figure 6.13. Device and connection configuration for two or more device output for
power scaling in the visible range...... 149
Figure 6.14. Total power output for single device, two devices together and three devices
under constant current of 300 mA...... 149
20 Figure A.1. Velmex program for laser doping of SiC. Dotted region shows the area to be
doped and each color shows the loop formed by the program and its progress for
covering the entire area...... 156
Figure B.1. Palladium dopant profile in laser doped 4H-SiC (undoped) substrate...... 158
Figure B.2. Boron dopant profile in laser doped 4H-SiC (undoped) substrate. The sample
has exceeded the solid solubility limit of 2.5x1020 cm-3 for B in SiC at the
surface...... 159
Figure B.3. SIMS depth profile for Al concentration in laser doped p-type Si...... 160
Figure B.6. GaP deep red LED fabricated by laser doping N in undoped GaP and
corressponding EL response of the LED ...... 162
Figure C.1. Schottky diode structure in 6H-SiC n-type wafer substrate...... 164
Figure C.2 I-V characteristics of the various contact metals and probes on 6H-SiC n-type
wafer substrate...... 165
Figure C.3. Change in the device temperatures with time under constant current
conditions...... 167
Figure C.4. Change in the total device resistance with time under constant current
conditions...... 167
Figure C.5. Cr contacts deposited on p-region of the white LED sample...... 168
Figure C.6 Degradation of Cr film under high current injection during EL measurements.
...... 168
Figure C.7. I-V characteristic of Contact 1 ...... 169
Figure C.8. I-V characteristic of Contact 2 ...... 169
Figure E.1 Debye temperature for various elements of the periodic table ...... 175
21 Figure E.2 Debye temperature for various compound semiconductors ...... 176
Figure F.1 Transmissivity and reflectivity of various dopants in undoped SiC...... 178
Figure F.2. Indirect to direct conversion by using dopants such as Cr, Se and Al in
undoped and 6H-SiC (n-type) substrate ...... 179
Figure F.3. Refractive index and absorption coefficient as a function of wavelength for n-
type 6H-SiC substrate (5x1018cm-3) for different currents at fixed 15V bias. .. 180
Figure G.1 Photocurrent measurement setup showing the light source, a pico-ammeter
and Cr and Al laser doped 6H-SiC (n-type) white light LED sample...... 182
Figure I.1. Reflectivity and transmissivity measurements for GE 124 Quartz substrate.189
Figure I.2. Temperature distribution at the quartz surface and depth for laser beam spot
size of 130 μm, average power of 10.5 W, pulse repetition rate of 5 kHz, pulse
on time of 90 ns and velocity of 1 mm/s ...... 190
Figure I.3. Change in the % absorptivity for laser Al and N doped and undoped GE 124
quartz samples...... 192
Figure J.1. Tunable FSS structure in Si using laser doping technique...... 196
Figure J.2. Results of the PMM based on the depletion region width and other device
parameters...... 197
Figure J.3. Mechanisms and concepts for obtaining a tunable FSS structure...... 199
22 LIST OF TABLES Table 1.1. Typical electronic and thermal properties of SiC and device fabrication issues
...... 41
Table 1.2. Dopants used for altering the properties of SiC...... 41
Table 1.3 Efficiencies and efficacies of various forms of commercially available lighting
in 2007. [Humphreys (2008)] ...... 45
Table 1.4. Total costs of ownership of a light bulb for one year and five years.
[Humphreys (2008)] ...... 45
Table 2.1 Physical properties of SiC substrates ...... 57
Table 2.2. Thermophysical and optical properties of SiC wafer substrates ...... 57
Table 2.3 Dopant precursors and their commercially available chemical name...... 59
Table 2.4. Nd:YAG (1064 nm) laser doping process parameters for SiC LED fabrication
...... 69
Table 3.1. Increase in the concentration beyond the solid solubility limit of various
dopants in silicon carbide by laser doping...... 78
Table 3.2 Experimental parameters for laser doping of N1, N2, N3 and blue LED...... 91
Table 3.3 Series resistances and carrier concentrations measured using I-V and C-V
respectively for the as-received substrate, 3 p+pn junction devices (N1, N2 and
N3) and p+pnn+ blue LED ...... 96
Table 4.1.Comparison of diffusion coefficient for various dopants in SiC under
conventional doping and laser doping...... 109
Table 5.1. Preliminary results on SiC White LED characteristics ...... 128
Table 6.1. Summary of SiC LED characteristics on 6H-SiC substrates with different
contact metal and contact configurations...... 131
23 Table 6.2. Summary of SiC LED characteristics on 4H-SiC substrates with different
contact metal and contact configurations...... 132
Table 6.4. Summary of measurements for color temperature tuning with two devices. 137
Table 6.5. Experimental parameters and light output characteristics for SiC device during
the excess injected current effect on SiC device output study...... 146
Table 6.6 Experimental parameters and light output characteristics for volume/area
scaling ...... 147
Table B.1. Experimental parameter for laser doping of Pd in undoped 4H-SiC ...... 158
Table B.2. Experimental parameter for laser doping of Pd in undoped 4H-SiC ...... 159
Table B.3. Experimental parameters for laser doping of Al in p-type Si ...... 160
Table B.4. Experimental parameters for laser doping of Al in p-type Si ...... 162
Table D1. Calculated resistance values from the I-V measurements for different contacts
...... 166
Table G.1. Photocurrent Measurements on C-1 (6H-SiC) and C-4 (4H-SiC) white light
LED samples and as received 4H-SiC (p-type) and 6H-SiC (n-type) silicon
carbide wafer samples...... 184
Table I.1. Physical and thermal properties of GE 124 Quartz substrates...... 189
Table I.2. Experimental parameters used during doping of n and p type dopants in GE
124 quartz substrate...... 191
Table I.3. Electronic and electrical properties of as received, nitrogen doped and
aluminum doped GE 124 quartz substrate...... 192
Table J.1. Parameter used for simulation of tunable FSS using PMM ...... 197
24 Table J.2. Laser doping parameter used for fabrication of Schottky diode for tunable FSS
...... 200
25 LIST OF SYMBOLS/ABBREVIATIONS
A Absorptivity (%)
2 Ai Area of the incident laser beam (cm )
Al Aluminum
AlGaAs Aluminum gallium arsenide
AlGaAsP Aluminum gallium arsenide phosphide
B Boron
BOE Buffered oxide etch
-3 C0 Surface concentration (cm )
C (z, t) Concentration at any depth z after time t (cm-3)
CFL Compact fluorescent lamps
C-V Capacitance-Voltage
CCT Correlated Color Temperature
CRI Color Rendering Index
Cr Chromium
D (T (z, t)) Diffusivity of Cr as a function of temperature
D0 Diffusion coefficient
DAP Donor-Acceptor Pair
DLTS Deep Level Transient Spectroscopy
e Electronic charge (C)
Em Electromotive force due to electromagnetic field of laser beam (V/cm)
Ey Elastic modulus of SiC (GPa)
El Electromagnetic field of the laser beam
26 Eg Energy gap (eV)
Ec Conduction band
Ev Valence band
EDS Energy dispersive spectroscopy
Er Erbium
EL Electroluminescence
Eu Europium
FIB Focussed ion-beam
H (t-tp) Heaviside step function
GaN Gallium Nitride
GaAsP Gallium arsenide phosphide
I Laser beam intensity (W/cm2)
IU Uniform laser beam irradiance
I-V Current-Voltage
InGaP Indium gallium phosphide
InGaN Indium gallium nitride
J Total mass flux
JF Fickian mass flux
Js Mass flux due to thermal stress
Jl Mass flux due to laser field
KB Boltzmann constant (J/K. mole)
LED Light Emitting Diode
27 lm lumens
Pd Palladium
N Nitrogen
Q Activation energy (eV)
R Reflectivity of SiC substrate (%)
Se Selenium
Si Silicon
SiC Silicon Carbide
SIMS Secondary Ion Mass Spectrometry
SSL Solid state lighting tp Laser pulse duration (ns) tb Laser substrate interaction time (ns)
T0 Initial temperature (K)
T (0.t) Surface temperature (K)
T (z, t) Temperature at any depth z after time t (K)
Tb Terbium
TEM Transmission electron microscopy
Th Thulium
TMA Trimethyl aluminum v Velocity of the substrate (cm/s)
WBG Wide Band Gap z Laser beam propagation direction (cm)
ZnO Zinc Oxide
28
Greek symbols
-1 αe Thermal expansion coefficient (K )
κ Thermal conductivity (W/m K-1)
ρ Density (gm/cm3)
σ Thermal stress (GPa)
ε Thermal strain
α Thermal diffusivity (cm2/s)
29 TECHNICAL PATENTS, PAPERS AND PRESENTATIONS
Patents
[1]. Sachin Bet and Aravinda Kar, “Deposition of Crystalline Layers on Polymer
Substrates Using Nanoparticles and Laser Nanoforming”, patent pending 2006.
Journal Publications:
[1]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Laser Doping of Chromium as a
Double Acceptor in Silicon Carbide with Minimum Crystalline Damage and
Maximum Ionized Dopants”, Acta Materilia, Vol. 56, pp. 1857-1867, 2008.
[2]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Effect of Laser Field and Thermal
Stress on Diffusion in Laser Doping of SiC”, Acta Materilia, Vol. 55, pp. 6816-6824,
2007.
[3]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Laser-Doping of Chromium in 6H-
SiC for White Light Emitting Diodes”, Journal of Laser Applications, Vol. 20, 1, pp.
1-7, 2008.
[4]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Laser-Doping of Silicon Carbide
for p-n Junction and LED Fabrication”, Physica Status Solidi (a), Vol. 204, Issue 4,
pp. 1147-1157 , 2007.
[5]. Sachin Bet and Aravinda Kar, “Thin film deposition on plastic substrates using
silicon nanoparticles and laser nanoforming”, Materials Science and Engineering B,
Volume 130, pp 228-236, 2006.
30 [6]. Sachin Bet and Aravinda Kar, “Laser Deposition of Silicon Films using Nanoparticle
Precursor”, Journal of Electronic Materials, Volume 35, Issue 5, pp 993-1004, 2006.
Conference Publications:
[1]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Laser Doping of Selenium and
Chromium in p-type 4H-SiC”, ICSCRM 2007.
[2]. N. Quick, S. Bet and A. Kar, “Laser Doping Fabrication of Energy Conversion
Devices”, Materials Science and Technology Conference, September, 2007.
[3]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Silicon Carbide White Light LEDs
for Solid-State Lighting”, SPIE Photonics West OPTO 2007, California, January,
2007.
[4]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Laser Fabrication Of Silicon
Carbide Light Emitting Diodes”, 25th International Congress on Applications of
Lasers & Electro-Optics, Arizona, November, 2006.
[5]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Laser-Patterned SiC Blue-Green
LED”, Materials Research Society Symposium B, San Franscisco, April, 2006.
[6]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Laser Thin Film Deposition on
Plastic Substrates Using Silicon Nanoparticles for Flexible Electronics”, 24th
International Congress on Applications of Lasers & Electro-Optics, Miami,
November, 2005.
[7]. Chong Zhang, Sachin Bet, Islam A.Salama2, Nathaniel R. Quick and Aravinda Kar,
“CO2 Laser Drilling of Microvias Using Diffractive Optics Techniques; I
.Mathematical Modeling”, Inter Pack 05, The ASME/Pacific Rim Technical
31 Conference on Integration and Packaging of MEMS, NEMS and Electronic Systems,
July 17-22 San Francisco, CA, 2005.
[8]. T. J. Mahaney, A.V. Muravjov, M.V. Dolguikh, T. A. Winningham, R. E. Peale,
Zhaoxu Tian, Sachin Bet, Aravinda Kar, and M. Klimov, “Laser doping of
germanium”, SPIE proceedings, Vol. 2, 5713A-12, January, 2005.
Oral Presentations:
[1]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Laser-Doped SiC Substrate-Based
Single-Chip White Light-Emitting Diodes with Improved Color Temperature and
Enhanced Power Output”, International Symposium on Semiconductor Light
Emitting Devices, TMS, April, 2008.
[2]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Tunable Optical Structures in
Silicon Carbide”, 26th International Congress on Applications of Lasers & Electro-
Optics, Florida, October, 2007.
[3]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Silicon Carbide White Light
Emitting Diodes”, TMS annual meeting, February, 2007.
[4]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Effect of Dopants in Silicon
Carbide for Solid-State Lighting”, Materials Research Society Symposium F, San
Franscisco, April, 2007.
[5]. Sachin Bet, Nathaniel Quick and Aravinda Kar. “Laser-Doped Silicon Carbide White
Light-Emitting Diodes with Multiple Dopants”, Graduate Research Forum, UCF,
Orlando, April, 2007.
32 [6]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Silicon Carbide White Light LEDs
for Solid-State Lighting”, SPIE Photonics West OPTO 2007, California, January,
2007.
[7]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Laser Fabrication Of Silicon
Carbide Light Emitting Diodes”, 25th International Congress on Applications of
Lasers & Electro-Optics, Arizona, November, 2006.
[8]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Laser-Patterned SiC Blue-Green
LED”, Materials Research Society Symposium B, San Franscisco, April, 2006.
[9]. Sachin Bet, Nathaniel Quick and Aravinda Kar, “Laser Thin Film Deposition on
Plastic Substrates Using Silicon Nanoparticles for Flexible Electronics”, 24th
International Congress on Applications of Lasers & Electro-Optics, Miami,
November, 2005.
[10]. Patent Committee Presentation, “Deposition of Crystalline Layers on Polymer
Substrates Using Nanoparticles and Laser Nanoforming”, January 2006.
Poster Presentations:
[1]. Laser Doping of Selenium and Chromium in p-type 4H-SiC, ICSCRM 2007.
[2]. Laser Doping Fabrication of Energy Conversion Devices, Materials Science and
Technology Conference, September, 2007.
[3]. Laser Doping Of Wide Bandgap Semiconductors: Fabrication of Laser Diodes and
Wireless Optical Sensors, CREOL Industrial Affiliates Day, May, 2006.
33 [4]. Novel dopants in Silicon Carbide for Light Emission, AVS Florida Chapter's 35th
Annual Symposium and Equipment Exhibition, University of Central Florida,
Orlando, Florida, March 8-11, 2007.
[5]. Laser Nanoforming of Polycrystalline Silicon Thin Film for Flexible Electronics,
CREOL Industrial Affiliates Day, May 1, 2005.
[6]. Polycrystalline Silicon Thin Film Using Silicon Nanoparticles For Flexible
Electronics, AVS Florida Chapter's 33rd Annual Symposium and Equipment
Exhibition, University of Central Florida, Orlando, Florida, March 13-18, 2005.
34 CHAPTER 1: INTRODUCTION
A light emitting diode (LED) is a semiconductor device which emits visible, infrared or
ultraviolet radiation by recombination of electron holes due to flow of electric current through it.
Essentially it is a p-n junction device with p- and n-regions made from the same or different
semiconductors. The color of the emitted light is determined by the energy of the photons, and in general, this energy is usually approximately equal to the energy band gap Eg of the semiconductor material in the active region of the LED. III-V semiconductors such as GaN,
GaAs, GaP, AlGaAs, InGaP, GaAsP, GaAsInP, AlInGaP, etc. are the common constituents of an
LED. However, which materials should be used for which LED depends on the band structure, choice of color, performance and cost. Typically, a semiconductor chip embodying an LED is
250×250 μm2 which is mounted on one of the electrical leads. The top of the chip is electrically connected to the other leads through a bound wire. The epoxy dome serves as a lens to focus the light and as a structural member to hold the device together. Operating currents at a forward voltage of about 2-3 V are usually in the range of 1-50 mA. [Schubert (2003)]
1.1 History of LED Development
The electroluminescence phenomenon was first discovered in 1907 by H. J.
Round from a SiC (carborundum). Although the discovery occurred nearly a century ago,
investigation related to this phenomenon did not take place until 1960. After the
invention of the maser (microwave amplification by stimulated emission of radiation) in
1954 by Townes and his collaborators and the subsequent operation of optical masers and
lasers in ruby, semiconductors were suggested for use as laser material. In 1960, Nick
Holonyak, who was working at General Electric at the time, started researching
35 compound semiconductor alloys. He and others had been working with gallium arsenide
(GaAs), a direct-bandgap semiconductor. It was known then that by inserting a p-n
junction into GaAs, infrared radiation was emitted. In 1962, the first GaAs1-xPx red laser
was made by Nick Holonyak’s group. General Electric sold the first commercial lasers
for $2600 and commercial LEDs for $260 in the same year. In 1967, J.M. Woodall, H.
Rupprecht, and G.D. Pettit fabricated efficient light emitting diodes from Ga1-xAlxAs.
High-volume production of GaAsP LEDs did not, however, begin until 1968 when the introduction of GaP:( Zn, O) LEDs was about to take place. Both GaAsP and GaP:(Zn,
O) LEDs exhibited an efficiency of about 0.1 lm/W and were available only in the color red. In late 1960's and early 1970's it was discovered, that nitrogen can provide an efficient recombination center in both GaP and GaAsP. The discovery led to the commercial introduction of red, orange, yellow, and green GaAsP:N and GaP:N LEDs with an improved performance in the range of 1 lm/W. Later it was found that both homo-structure AlGaAs and heterostructure AlGaAs LEDs could offer potential performance advantages over GaAsP and GaP homo-junction LEDs. Nevertheless, it was not easy to realize it in practice because more than a decade of development was needed to realize high-volume liquid-phase-epitaxy (LPE) reactors capable of growing high- quality multilayered device structures. As a result, such LEDs did not become commercially available until the early 1980's. The performance of these LEDs was, however, significantly improved, which ranged now from 2 to 10 lm/W depending upon the structure employed. Thus, for the first time, LEDs broke the efficiency barrier of filtered incandescent bulbs, enabling them to replace light bulbs in many outdoor lighting applications.
36
2006 160 Lumens/Watt White LED (GaN- based) 100 Lumens /Watt OLED
Figure 1.1. History of LED Development [Press release (2006)]
As a result of continuous efforts in research, AlInGaP based orange and yellow LEDs
with efficiencies higher than 10 lm/W were developed in early 1990's. Interestingly,
conventional techniques such as LPE or halide-transport vapor phase epitaxy (HVPE)
proved intractable for the growth of these LEDs. The metal-organic-vapor-phase epitaxy
(MOVPE) system which emerged as a powerful crystal-growth technique in the late
1960's and yielded high-performance AlGaAs LEDs in the late 1970's was eventually employed for a more controlled growth of heterostructure AlInGaP LEDs. The performance of these LEDs was satisfactory. In early 1990's SiC based LEDs emerged as potential blue color emitters. The development of the brighter blue LEDs, using ZnSe, which have shown promise in the research laboratory, is emerging as a major area in the
1990's. Figure 1 shows the history of the LED development and confirms that LED’s
37 have broken all the barriers in terms of performance 2006. [Schubert (2003), Berch et. al
(1992), Morkoc et. al. (1997), Crawford (1992)].
Parallel to the development of visible LEDs infrared (IR) LEDs also attracted interests. Instead of being detected by eyes, these LEDs are detected by photodiodes or phototransistors. Therefore, IR LEDs can function as important tools for transmitting data. How fast these data will be transmitted depends on the switching speed of the IR
LEDs. Therefore, this switching speed is quite unimportant for visible LEDS, this is an important performance parameter for IR LEDs.
1.2 Wide bandgap (WBG) Semiconductors
WBG semiconductors are a class of materials with energy gap greater than 2.0
eV. The WBG materials of interest to the microelectronic industry can be broadly
classified into two groups based on their applications. The first class includes the
materials for light emitting applications where the main requirement is the direct bandgap
characteristic that allow for an efficient light (Tunable multicolor and White). This class includes gallium nitride (GaN), and the various II-VI compounds such as the tellurides, sellenides, sulfides and oxides of zinc (Zn), cadmium (Cd), and mercury (Hg). The other group includes the WBG materials for high power and high temperature applications that includes silicon carbide (SiC), aluminum nitride (AlN), boron nitride (BN), diamond (C) and diamond-like carbon (DLC). The requirements for this class of applications are; wide bandgap, indirect band characteristics and excellent thermal properties. WBG semiconductor p-n junction devices have potential for operating temperatures greater than
3000°C, as shown in Figure 2. [Pearton (2000), Levinshtein et. al. (2001)]
38
Figure 1.2. Operating temperature of p-n junctions for semiconductors as a function of bandgap [Pearton (2000)].
1.2.1 Silicon Carbide
Silicon carbide is the only tetrahedral compound known to be formed by the IV-B group elements in the periodic table. SiC exhibits a one-dimensional type of polymorphism as do many other close-packed materials. While SiC crystallizes mainly in three lattice structures (cubic, hexagonal and rhombohedral), more than 200 different polytypes are known to exist for SiC with the exact physical properties of each type being dependent on the crystal structure. The polytypes are similar within the closest-packed planes (basal planes) and are different in the stacking direction normal to these planes.
The stacking sequence, expressed by the ABC notation, describes the alternative arrangements of the different atomic layer. Here, each letter represents a bilayer of individual Si and C atoms. The ABCA stacking sequence resulted in the zinc blend crystal structure where each SiC bilayer can be oriented only into three different positions with respect to the lattice point while maintaining the sp3-bonding configuration of the
39 SiC. This particular arrangement is known as 3C-SiC or the β-phase. When the stacking
sequence is ABA, the crystallographic symmetry is hexagonal (Wurtzite) and the material
is known as 2H-SiC. All other polytypes are intermixed forms of both the hexagonal and
the cubic symmetries. [Simon (1970), Harris (1995), Levinshtein et. al. (2001) and Feng
et.al. (2004)]. Two most common polytypes of the hexagonal category are 4H-SiC and
6H-SiC where the overall symmetry is hexagonal with stacking sequence of
ABCBABCB and ABCACBABCACB, respectively. The 6H-SiC composed of one-third hexagonal bonds and two-third cubic bonds while the 4H-SiC has an equal contribution of the hexagonal and cubic bonding configurations. All the non-cubic polytypes are collectively known as α-SiC. [Simon (1970), Harris (1995)]
Different polytypes of silicon carbide have different physical and electronic properties. At the room temperature, 3C-SiC has the lowest bandgap energy (Eg ≈ 2.4 eV)
with very high electron mobility second to 4H-SiC (Eg ≈ 3.02 eV).. The ratio of the
electrons mobility in the basal plane (μ//) to that along the normal to these planes (μ⊥) is
about 1.25 and 0.16 for 4H-SiC and 6H-SiC polytypes, respectively. Due to this anisotropic property of 6H-SiC, 4H-SiC polytype is the first choice in microelectronic device applications [Pensel et. al. (1993), Neudeck (2000)].
A proper understanding of the diffusion mechanism for various species in SiC and the corresponding optical response of the doped substrate will lead to the fabrication
LEDs. It exhibits luminescence due to transitions between impurities (e. g., between nitrogen and aluminum levels) and between free and bound exciton states. Typical properties of SiC which are superior to other WBG’s and the inherent drawbacks
40 Table 1.1. Typical electronic and thermal properties of SiC and device fabrication issues
Silicon Carbide 3C-SiC 4H-SiC 6H-SiC a/c=0.3 a/c=0.2 Bandgap (eV) (Indirect) 2.36 3.23 3.00 Thermal conductivity (W/mK) 350 370 490 Thermal diffusivity (cm2/s) 1.6 1.7 2.2 Mobility(cm2/Vs) Electron 800 900 400 Hole 300 120 90 Junction temperature 600-650°C Break down field (MV/cm) 5 Device structure Homojunction Light emission Single Chip (DAP recombination mechanism) Doping issues Conventional doping difficult
regarding SiC device fabrication are listed in Table 1.1. The influence of dopants on the properties of SiC is listed in Table 1.2. LED’s, pin diodes, Schottky diodes, photodiodes and other semiconductor devices can be fabrication using specific dopants. [Harris
(1995)]
Table 1.2. Dopants used for altering the properties of SiC.
Dopants Substrate Product N, P, Se SiC n-type semiconductor Al, B, Cr SiC p-type semiconductor Al, N 4H-SiC, Cr coupler Violet LED Al, N 6H-SiC Blue LED B 6H-SiC Yellow LED B n-type 4H-SiC Green LED Al n-type SiC UV Photodiode N p-type SiC UV Photodiode - 6H-SiC or 4H-SiC, thin Cr layer Schottky diodes
41 1.2.2 Silicon Carbide LED Development
SiC is an indirect bandgap semiconductor (in which both the momentum and energy of an electron needs to be changed to move it from valence band to conduction band) whereas GaN, GaAs are direct bandgap semiconductors in which only the energy of an electron needs to be changed to move it from valence band to conduction band.
[Pankove (1971)]. Therefore, SiC LEDs are generally less efficient than GaN and GaAs based LEDs. While LEDs are based on spontaneous emission, they are usually made of a p-type region, n-type region and an active layer, which is usually formed by the bulk material, between the two doped regions. The active layers of normal LEDs are intentionally undoped or lightly doped in order to increase the light output power. Hence, the dopant concentration plays an important role in the output power of a LED. [Brander
(1972)]. The main light producing mechanisms in silicon carbide have been identified to be as shown in Figure 1.3:
Ec (a) (b) (c) (d) (b)
Ev
Figure 1.3. Schematic representation of various recombination processes (a) free electron-hole recombination, (b) recombination of donor, acceptors or isoelectronic traps
(c) recombination of two bound carriers and (d) recombination within a localized center
42 Though phenomenon of EL was discovered in SiC in 1907, however LED fabrication in
SiC began in late 1960’s. SiC films were prepared by more careful processes [Violin et.al
(1969)], and p-n junction devices were fabricated, leading to blue light emitting diodes.
[Brander et al.(1969)] were the first to fabricate a blue SiC LED and since then other
groups have followed suit.[Potter et. al (1969), Ikeda et. al. (1979), Hoffman et. al.
(1982), Dmitriev et. al.(1986, 1989), Vishnevskaya et. al. (1990), Suzuki et. al. (1991),
Rahman et. al. (1992), Edmond et. al. (1995, 1997), Vlaskina (2002), Kamiyama (2006)].
Electrical to optical conversion efficiencies were only 0.005% [Potter et. al. (1969)]. Cree
Research [Edmond et. al. (1995, 1997)] presently markets the best performing GaN blue
LEDs on SiC substrates, which have a pure blue emission centered at 470 nm. Groups at
Siemens AG [Hoffman et. al. (1982)], Sanyo and Sharp [Ikeda et. al. (1979) and Suzuki
et. al. (1991)] have also developed prototype devices and some have reached the
marketplace. Cree's devices are reported to radiate 18.3 μW at a 25 mA (3 V) forward
bias with a spectral half width of 69 nm. Due to the indirect bandgap the efficiency of
these devices is only 0.02 - 0.03 %, but that is partly compensated for by the ability to
drive the SiC LEDs at higher currents. At 50 mA, 36 μW of output has been achieved.
LEDs run at 50 mA show a typical degradation of 10 -15% over 10,000 hours,
significantly less than GaP LEDs, and once again attesting to the durability of SiC.
Additionally, Vlaskina (2002) demonstrated a SiC green LED having the same brightness
as CREE’s (Cree Research, Inc.) GaN-based LEDs, but higher stability and reliability
(e.g., repeatability and constant brightness) over a wide temperature range, simpler
device design (without the multilayer AlGaN or GaN quantum well structures). Edmond
43 et al. (1995, 1997) fabricated blue LEDs, green LEDs and UV (Ultraviolet) photodiodes
using SiC. Dmitriev et. al. (1986, 1989) showed violet, green and red LEDs.
1.3 Motivation for white LED
An incandescent light bulb uses electricity to heat a coiled tungsten wire in an
evacuated glass bulb. The temperature of the wire is about 3,500 K and it glows white-
hot, radiating white light. The lifetime of an incandescent light bulb is typically 1000 h.
The spectrum of radiation emitted is very broad. It fills the entire wavelength range (400–
700 nm) of the human eye. Because its visible spectrum is so broad, it renders colors extremely well. An incandescent light bulb has a CRI approaching 100. However, it also emits strongly in the infrared as heat, beyond the response of the human eye. Only about
5% of the input electrical energy is converted to visible light, and the rest is emitted as heat.
Fluorescent tubes have much longer lifetimes (7,500– 30,000 h) than incandescent light bulbs (1000 h). The quality of white light emitted depends on the phosphors used, but a warm white fluorescent tube has phosphors emitting in the blue, green, and red ranges. The light conversion efficiency is typically 25%.
Compact fluorescent lamps (CFLs) usually consist of two, four, or six small fluorescent tubes, which can be straight or coiled. Their efficiency is typically 20% with an expected lifetime of ~10000 hrs. CFLs are likely to be a stop-gap measure to replace incandescent lamps, lasting until we have a more efficient, nontoxic source of white light at a reasonable cost. [Humphreys (2008)]
44 Solid-state lighting (SSL) is an emerging technology with the potential to surpass these luminous efficacy limitations, and at the same time to introduce new functionalities and designs in lighting. Based on semiconductor light emitting diodes, SSL has made remarkable progress in the past decade [Phillips et. al. (2007)], to the point where it is
Table 1.3 Efficiencies and efficacies of various forms of commercially available lighting in 2007. [Humphreys (2008)]
Type of Light Source Efficiency (%) Efficacy (lm/W) Incandescent light bulb 5 15 Long fluorescent tube 25 80 Compact fluorescent lamp (CFL) 20 60 High-power white LEDs 30 100 Low-power white LEDs 50 150 Sodium lamp (high-pressure) 45 130 White LED (10 year target) 60 200
Table 1.4. Total costs of ownership of a light bulb for one year and five years.
[Humphreys (2008)]
United States United Kingdom Type of Light Source 1 year 5 years 1 year 5 years 60 W Incandescent $18 $90 $36 $180 Compact fluorescent lamp (CFL) $8 $28 $11 $50 60 lm/W warm white LED $18 $36 $23 $58
Note: Values calculate assuming light bulbs used for 8 h/day at electricity cost of $0.1/kWh (U.S.) and $0.2/kWh (U.K.), cost of the incandescent bulb $0.5, cost of CFL $2 and cost of LED $14
45
now is very competitive with incandescent technology. There is much research and
development worldwide aimed at making SSL competitive with fluorescent technologies
in the coming decade, with ultimate target efficiencies in the 50% range.
A further benefit is that SSL does not contain toxic materials, whereas the
mercury vapor contained in fluorescent lamps is increasingly causing concern, to the point where used fluorescent lamps must be treated as hazardous waste in many areas.
Table 1.3 shows the efficiencies and efficacies of various forms of commercially available lighting in 2007.
It is also important to compare a cost structure over the period of lifetime of these sources. Table 1.4 gives the total costs of ownership of a light bulb for one year and five years.
1.4 Human Eye Response
The three curves in the Figure 1.4 above shows the normalized response of an
average human eye to various amounts of ambient light. The shift in sensitivity occurs
because two types of photoreceptors called cones and rods are responsible for the eye's response to light. The curve on the right shows the eye's response under normal lighting conditions and this is called the photopic response. The cones respond to light under these conditions. As mentioned previously, cones are composed of three different photo pigments that enable color perception. This curve peaks at 555 nanometers, which means that under normal lighting conditions, the eye is most sensitive to a yellowish-green
46
Figure 1.4. Human Eye’s response to light [Robinson et. al. (1984)] color. When the light levels drop to near total darkness, the response of the eye changes significantly as shown by the scotopic response curve on the left. At this level of light, the rods are most active and the human eye is more sensitive to the light present, and less sensitive to the range of color. Rods are highly sensitive to light but are comprised of a single photo pigment, which accounts for the loss in ability to discriminate color. At this very low light level, sensitivity to blue, violet, and ultraviolet is increased, but sensitivity to yellow and red is reduced. The heavier curve in the middle represents the eye's response at the ambient light level found in a typical inspection booth. This curve peaks at 550 nanometers, which means the eye is most sensitive to yellowish-green color at this light level. [Robinson et. al. (1984)].
1.5 Color Rendering Index (CRI)
An important characteristic is the ability of the white light to render the colors of objects in the environment to the human visual perception system in an accurate and
47 pleasing manner. The color rendering index is an internationally accepted measure of
how well a light source renders colors. The CRI varies between 0 and 100, with 100
representing perfect color rendering. A CRI of 90% is considered to be excellent, and the
maximum possible efficacy of a white light source with a CRI of 90% is 408 lm/W.
Light composed of just 550 nm wavelength would have a very high luminous efficacy of
radiation (683 lm/W) but would render well only objects that reflect at that wavelength.
To render well objects that reflect at other wavelengths, a light source must include light
that spans a wider wavelength spectrum, but that the human eye is less sensitive to.
Hence, the wider the included spectrum, the better the color rendering, but also the lower
the luminous efficacy. Such a CRI is considered excellent and would satisfy virtually all
white-light applications. Here, everyone use the internationally agreed upon metric for
the color rendering index [CIE (1932)]. In general, the higher the CRI, the lower the possible luminous efficacy. SiC based white LEDs emit a broad band spectrum thus
rendering excellent CRI. [Bet et. al. (2008)]
1.6 Correlated Color Temperature (CCT)
"Correlated Color temperature" is used loosely to mean "white balance" or "white
point". It is a characteristic of visible light determined by comparing its chromaticity with
a theoretical, heated black-body radiator. The temperature (in kelvin) at which the heated
black-body radiator matches the color of the light source is that source's color
temperature. The CCT for bright midday sun is 5200 K and average daylight 5500 K,
while the reference for pure white light is taken as 5500 K.
48 The CCT and CRI has always been an issue with the white light emitting diodes.
Purely GaN based white light LEDs are more towards the blue side due higher contribution towards the blue from the material itself and lower red contributing component. While the combination of GaN with phosphors are devoid of the deep green component. Purely phosphors based LEDs shows a good CRI but are extremely low in efficiency. While the RGB individual color emitting GaN LEDs based show excellent
CCT and CRI, however these individual colored LEDs respond differently to the drive current, operating temperature, dimming and operating time. Also additional controls are needed for color consistency which further adds to the cost.
1.7 Current white LED technology based on direct bandgap semiconductors
Currently there are a number of ways to use GaN-based LEDs to make white light:
1.7.1 Blue LED and yellow phosphor
As stated previously, nearly all white LEDs sold today use a blue GaN/InGaN
LED plus a yellow phosphor. The blue LED chip is covered with a thin layer of a phosphor that emits yellow light when excited by the blue light. The phosphor layer is sufficiently thin that some blue light is transmitted through it, and the combination of blue and yellow produces a cool white light.
1.7.2 Red plus green plus blue LEDs
This method, mixing red, green, and blue (RGB) LEDs is the obvious way to produce white light. However, this approach has three basic problems. The first is that the efficiency of green LEDs is much less than that of red and blue LEDs, for reasons that are not yet understood (this is known as the “green gap” problem). Hence, the overall
49 efficiency of this method is limited by the low efficiency of the green. Second, the
efficiencies of red, green, and blue LEDs change over time at different rates. Hence, if a
high-quality white light is produced initially, over time, the quality of the white light
degrades. However, this process is slow and can be corrected using automatic feedback.
Third, because the emission peaks of LEDs are narrower than those of most phosphors, red plus green plus blue LEDs will give a poorer color rendering than red plus green plus blue phosphors. This problem can be minimized by a careful choice of LED emission wavelengths, and of course, more than three different color LEDs can be used for better coverage of the visible spectrum. In particular, using four LEDs red, yellow, green, and blue can give a good color rendering.
1.7.3 Red, green, and blue quantum dots in a single LED
It is possible to produce a single LED with quantum dots of InGaN of different
sizes and compositions so that white light is emitted. This is a recent development, and
the efficiency, reproducibility, and lifetime of these LEDs are not yet known.
1.7.4 Near-UV or blue LED plus red, green, and blue phosphors
As already discussed, blue LEDs covered with yellow phosphors give a rather
cool white light. This is fine for many applications (e.g., displays, lighting in cars, buses,
yachts, key-rings, and cell phones), but the quality of light is probably not good enough
for home lighting, for which a warmer white light containing some red light is desirable.
Such warm white LEDs (blue LEDs plus yellow and red phosphors) are available
commercially now. However, the efficiency with which existing red phosphors are
excited using blue light is much less than that using near-UV light; hence, a better route
to higher quality white light might be to use a near-UV LED plus red, green, and blue
50 phosphors. There are no dangers in using a near-UV LED as thick phosphor layers would be used so that no near-UV light would be transmitted, in much the same way as the phosphor coating on fluorescent tubes and CFLs prevents the transmission of UV light.
[Humphreys (2008)]
However, there are many factors that the hindering these SSL technologies: efficiency, heat management, color rendering, lifetime, complexity of fabrication, toxicity and cost. Additionally most of the phosphors currently in use have been developed for the use with fluorescent tubes or CFLs that emit UV radiation, and hence, they have not been optimized for use with LEDs emitting in the visible spectrum. Thus, there is still thrust for fabrication of single chip, simple design, environmental friendly (free of toxic elements) and cheap LED with very good efficiency, lifetime and heat management. SiC to a certain extent satisfies most of these requirements.
1.8 Indirect bandgap semiconductor (SiC) for white LED
SiC is an indirect bandgap semiconductor, therefore, SiC LEDs are generally less efficient than direct band gap GaN and GaAs based LEDs. However, recent work by Pan
J. (2006) on use of deep-level transitions in semiconductor devices shows that indirect bandgap semiconductors can behave similar to direct bandgap semiconductors. There has been tremendous improvement in the quality of the SiC substrate post the stoppage of its use in late 90’s after evolution of GaN. The micropipes, dislocations and other intrinsic defects which were the prime hurdles for SiC technology are completely eliminated. The wafer growth technology has also advanced significantly wherein the capabilities are available to grow defect free 100 mm wafers. Doping issues with SiC has been
51 eliminated to a certain extent. Epilayer doping is still widely used however with advent of laser doping almost all the issues have been resolved. It is easy to incorporate both p-and n-type dopants with complete activation. Due its wide band gap and DAP mechanism for light emission, it is possible to obtained white LEDs which are single chip, simple structured and free from any toxic elements such as phosphors or Hg or Na used in other technologies. A revolution similar to GaP can be visualized in the SiC LED technology with all these developments. Cost which is primarily considered as a limiting factor in the advancement of any technology should not be a concern as the existing successful and popular GaN technology still utilizes SiC as substrate. This research aims at reestablishing the SiC LED Technology for environmental friendly white LED fabrication. Initial results in the development have lead to white LED with excellent color rending properties due its broad band emission along with a respectable power output.
52 1.9 Objectives
The purpose of this research was to fabricate pure green light emitting diodes in silicon carbide using laser doping. However with the initial encouraging results the research goals were extended further towards fabrication of white LEDs in SiC with enhanced power output in the visible range (~1 mW) and good color temperature (~5500 K). To achieve this goal following research accomplishments were desired and attained:
1. Defect free laser doping technique for incorporation of p and n-type dopants
2. Selection and use of novel dopant materials such as Cr, Se, B for tailoring light
emission properties
3. Development of diffusion model to understand the laser doping process by taking
into effect the concentration gradient, localized thermal stress and electromotive
force of laser beam
4. Study of activation of the incorporated dopants and their impurity levels for light
emission properties
5. Laser Fabrication of white light emitting diodes in SiC
6. Demonstrate an approach for color tuning, color mixing and power scaling.
53 1.10 Technical approach to achieve the objectives
Silicon Carbide LEDs
Laser Doping
Dopant diffusion p and n-type doping p and n-type dopant mechanism activation
Dopant defect levels Donor-Acceptor pair recombination mechanism
Al-N, Ga-N, Al-Se B-N, Cr-N,Mg-N,Cr-Se Al-Cr, Ga-Cr Zn, Cr, N, Se Violet-Blue Green-Yellow Orange Red
Efficiency White LEDs Color Temperature 7-9 nW 5500K
Indirect Low Low density Effective Free Extraction bandgap absorption of states of injection of carrier losses coefficient dopant atoms electrons for absorption at deep levels recombination losses Contacts Top device Dopant
Cr-Se, Phonon Doping Double donor / Ni, Al, Cu, Cu- Dopant Conc. Sn Debye Temperature acceptor Au, Cr, In, laser Contact Metal doping (Zn, Mg, Ga) Two or multi metallization Two Devices device approach
White LED Single Device Phonon -tail Enhanced High Voltage 3.66 mW, 0 K disappearance Power output High injection Efficiency: 0.25% Current
54 CHAPTER 2: LASER DOPING OF SILICON CARBIDE
2.1 Laser interaction with silicon carbide
Laser interaction with SiC is mainly dependent upon the following factors:
1. Material (SiC) : Thermo-physical and electronic properties, Surface condition,
Surface reactivity
2. Thermal Source (Laser beam) characteristics: output power and pulse energy,
spatial beam parameters such as shape and extent, temporal parameters such as
pulse duration and spatial and temporal coherence and mode.
3. Interaction (Doping): Laser scanning speed, irradiation ambient, beam delivery
optics, spot size or laser fluence, ambient atmosphere
Depending on the temperature and the ambient atmosphere during interaction one can
perform different processes in SiC such as doping, direct writing of conductive tracks,
nanostructure fabrication, annealing, surface modification, ablation and oxidation.
Therefore it is very important to estimate the temperature prior to performing each of
these processes.
Higher the temperature, higher is the diffusivity of the dopants and greater will be
the penetration depth for dopants. However, in case of LISPD it is very important to take
into consideration the phase transitions that may occur during the laser interaction. Figure
2.1 shows the phase diagram of SiC. From the phase diagram one can observed that the
highest temperature one can use prior to observing any phase transitions is 2830°C (3103
K). Therefore, prior to performing laser doping, a thermal model is developed and
presented in the following to estimate the laser processing parameters taking to effect all
the properties of the laser beam, SiC and their interaction. The optimum parameters for
55 laser doping were estimated for both 4H and 6H-SiC polytypes with preexisting dopants.
The results of this model were further utilized to develop a diffusion model for estimating diffusivity of various dopants in SiC.
Figure 2.1 Phase diagram in Si-C the system. α is a solid solution of C in Si and β is a
solid solution of Si in C. [Tairov et. al. (1988)]
2.1.1 Material (SiC)
2.1.1.1 Physical properties of SiC substrates
Two different polytypes (6H and 4H) and three different wafer substrates
depending upon the existing dopant were use in this study. Silicon carbide wafers are
usually pre-doped with Al (p-type) or N (n-type) as a base dopant. The corresponding
properties of SiC substrates are listed in Table 2.1.
56 Table 2.1 Physical properties of SiC substrates
Wafer Thick Crystal Surface Dia. Carrier Dopant conc. roughness Polish Substrate (μm) Structure (inch) conc. (cm-3) (atoms/cm3) [Å]
6H-SiC (n- Both 17 18 444 Hexagonal 1.92 2 3.4x10 5x10 type-N) sides
4H-SiC (p- one 17 19 454 Hexagonal 1.72 2 3x10 1x10 type-Al) side
2.1.1.2 Thermophysical and optical properties
Prior to laser doping, all the optical properties were measured for both the wafer
substrates. Thermophysical properties were obtained from Harris (1995). These
properties are listed in Table 2.2. The thermal properties such as thermal conductivity and
thermal diffusivity along with the absorptivity are utilized in the thermal model to
estimate the temperature at the wafer surface. Electronic properties such as dopant
concentration and corresponding carrier concentration assists in understanding the dopant
behaviour post laser doping of new dopants and their activation. Electrical properties are
useful during the device design and in estimating of device operating parameters.
Table 2.2. Thermophysical and optical properties of SiC wafer substrates
Sample Thermal Thermal Thermal Young’s Absorptivity Density conductivity Diffusivity expansion modulus at 1064 nm [gm/cm3] [W/cm.K] [cm2/s] coefficient [GPa] [%] [K-1] 6H-SiC 4.9 2.2 4.6×10-6 392 13.8 3.21 (n-type) 4H-SiC 3.7 1.7 4.6×10-6 392 70.3 3.21 (p-type)
57 Additionally optical properties allow the selection of laser source for a particular device design and in present work a Q-switched Nd:YAG pulsed laser (1064 nm wavelength) was utilized to perform the doping.
2.1.2 Dopant precursors
The dopant precursors can be either in solid, liquid or gaseous form depending on
the laser doping technique. In the current study, solid and gaseous forms of dopant
atmospheres are used. An organometallic compound is usually available for most of these
dopants which can serve as a very good gaseous source. This organometallic compound
is heated in a water bath and a carrier gas (inert gas e.g. Ar) is usually used to transport
the vapor to the doping chamber. A mass flow controller or a pressure controller can be
used to control the dopant atmosphere concentration. Except for boron and rare earth
elements all the other elements are in organometallic compounds. The detailed procedure
for use of these dopants during laser doping has been described in the laser doping
section. Table 2.3 shows the dopant precursors that were used in the present study.
2.1.3 Laser characteristics for doping
Two different laser sources Nd:YAG (1064 nm) and Excimer laser (193 nm, 248 nm and 353 nm) were used. However most of the work was carried out with Nd:YAG
due to low photon energy and greater penetration depth.
1. Lamp-pumped Nd:YAG source (LEE-8150MQ) irradiating laser of wavelength
(λ) = 1064 nm. This laser source runs in both the continuous (CW) and the pulsed
(Q-switched) mode. In the CW mode the out put power of the laser varies in the
58 Table 2.3 Dopant precursors and their commercially available chemical name.
Dopant Element Chemical Name
Aluminum Trimethyl Aluminum (CH3)3Al 25% w/w in hexane
Boron Boron (B) Powder Crystalline 325 mesh 98% (metal basis)
Triethylboron B(C2H5)3
Nitrogen Ultrahigh pure nitrogen (N) 99.999% purity
Chromium Bis (ethylbenzene) Chromium (C2H5)XC6H6-xCr where x = 0-4
Selenium Diethyl Selenide Se(C2H5)2
Zinc Diethyl Zinc Zn(C2H5)3
Magnesium Bis(ethylcyclopentadienyl)magnesium
Gallium Triethyl Gallium Ga(C2H5)3
Europium Europium 2,2,6,6, Tetramethyl 3,5 heptanedionate (C33H57O6Eu)
Thulium Thulium 2,2,6,6, Tetramethyl 3,5 heptanedionate (C33H57O6Tm)
Terbium Terbium 2,2,6,6-Tetramethyl 3,5-Heptanedionate (C33H57O6Tb)
Erbium Erbium 8 – Hydroxyquinolinate (C27H18N3O3Er)
range of 8 to 200 W. In the pulsed mode the characteristics of the laser irradiation can be tuned in the following ranges; pulse on-time 60 - 240 ns, pulse energy 0.6-3.0 mJ, and pulse repetition rate up to 40 kHz. It emits a multimode laser beam with a nominal diameter of 6 mm. The laser beam delivery system consists of apertures (0.5 mm-3 mm), bending mirror and a focusing lens (25 mm -150 mm).
2. Multigas Excimer laser (LPX 200i) with three lasing media; ArF with wavelength
(λ) = 193 nm, KrF with wavelength (λ) = 248 nm, and XeF with wavelength (λ) =
59 351 nm. For all three wavelengths; the maximum repetition rate is 100 Hz, the
nominal pulse duration is 20 ns and the pulse-to-pulse stability is ± 5 %.
2.1.4 Doping Methods
2.1.4.1 Conventional doping (ion-implantation, thermal diffusion)
SiC device fabrication has faced several technological difficulties particularly with doping. Defect generation during doping, inherent materials defects and unstable contacts for high temperature applications are some of the other issue with SiC. Ion implantation and furnace annealing that are commonly employed in microelectronic industries to incorporate and activate dopants are not readily adaptable to SiC technology due to the low diffusion coefficients of dopants in SiC (Figure 2.2). It has been observed that ion implantation modifies the stoichiometry, introduces defects and changes the homogeneity of the substrate [Pearton (2000), Zetterling et. al. (2002)]. Ion-implanted substrates require extremely high annealing temperatures (1600-2200°C), where the entire wafer is isothermally heated to a predetermined annealing or drive-in temperature.
This process can cause undesirable dopant diffusion into any previously prepared sub- layers. Photolithographic patterning is necessary to define the areas across the sample to be selectively doped for conventional furnace annealing. This usually requires up to 10-
15 individual processing steps. For fabricating ultra shallow n+-p and p+-n junctions, ion implantation capabilities require 10-20 keV energy implants with significant planer or axial ionic channelling imposing severe constraints on the entire process [Pearton (2000),
Levinshtein et.al. (2001) and Feng et. al. (2004)].
60
Figure 2.2. Temperature dependence of the diffusion coefficient for various dopants in
SiC [Salama (2003)].
2.1.4.2 Laser doping
The use of high power lasers with nanosecond pulses enables deposition of a large amount of energy in a very short time onto the near surface region of a substrate, while maintaining the rest of the substrate at room temperature. The laser heating of the substrate in the presence of a dopant atmosphere (gas, solid or liquid phase) causes the diffusion of dopants under concentration gradients. Laser doping can be classified as laser thermal processing (LTP), gas immersion laser doping (GILD) and laser-induced solid-phase doping (LISPD) [Salama (2003)] depending on the dopant source.
Direct laser doping is a single step process and it enables creation of very shallow dopant profiles with concentrations up to the solid solubility limit. Pulsed lasers are mainly used for large area (sheet) doping, while continuous wave (CW) lasers allow
61 direct writing. [Salama (2002) and (2004)] developed and optimized the laser doping technique to successfully incorporate both n-type (nitrogen) and p-type (aluminium) dopants into SiC substrates. [Tian (2005) and (2006)] utilized this technique for fabrication of Schottky and pin diodes and optical structures in SiC. The laser doping technique is employed in this work for fabricating SiC p-n junction diodes that have been found to operate as a light-emitting diode (LED).
2.2 Thermal model for selection of laser doping parameters
The temperature distribution in the substrate generally affects the driving forces
for the diffusion of dopant atoms in the wafer whose surface, which is at z = 0 , was irradiated with a Nd:YAG Gaussian laser beam propagating in the z direction (Figure
2.5). Although the irradiance of a Gaussian beam varies radially, a uniform irradiance is considered in this study to develop a simple one-dimensional thermal model to calculate the temperature distribution. Actually laser beams with uniform irradiance profile would be necessary for doping applications to achieve uniform dopant depths across the doped region. To simplify the thermal model, the absorptivity and thermo-physical properties of the substrate are also considered constant. The laser irradiance and the pulse repetition rate were varied to calculate the optimum temperature for laser doping and to select the laser doping parameters such that the temperature of the wafer remains below the peritectic transition temperature (3100 K for SiC) to prevent any crystalline phase transformation in the wafer.
62 3200 Peritectic Temperature
3000 4H-SiC (p-type) 6H-SiC (n-type) 2800
2600
2400 Temperature (K) Temperature
2200
0 200 400 600 800 1000 1200 Depth (nm)
Figure 2.3 Transient temperature distributions along the depths of n-type 6H-SiC and p- type 4H-SiC wafers for laser irradiation times of t = 64 and 89 ns respectively.
The governing equation for laser heating of the wafer, which was considered infinite in the x and y directions and semi-infinite in the z direction, is given by
∂ )t,z(T ∂ 2 )t,z(T = α (2.1) ∂t ∂z 2 with the boundary and initial conditions defined as
∂T − k []1 −−= )tH(tAI at z = 0, 2.2(a) ∂t U p
(), =∞ TtT 0 , 2.2(b) and
()0, = TzT 0 , 2.2(c)
63 Where T (z, t) is the temperature of the wafer at any depth z at time t. α, k and A are the thermal diffusivity, thermal conductivity and absorptivity of the wafer respectively. H (t
– tp) is the Heaviside step function, tp is the laser pulse-on time and T0 is the initial
temperature of the wafer. IU is the uniform laser irradiance incident on the wafer surface,
P0 which is given by IU = 2 , where P0 is the average laser power, pr is the pulse πrtp 0pr
repetition rate and r0 is the radius of defocused laser beam at the wafer surface. The focal
plane, i.e., the beam waist was 10 mm below the top surface (z = 0) of the wafer.
This thermal model can be solved using the LaPlace transform [Tian et. al. (2005)
and (2006)] to obtain the following solutions:
2 U AI α ),( TtzT += []ierfct η)( for t < tp (2.3) 0 k
and
⎡ ⎡ ⎛ ⎞⎤⎤ 2 U AI α ⎜ t ),( TtzT += ⎢ []ierfct η)( −− ⎢ierfctt η ⎟⎥⎥ for t ≥ tp, (2.4) 0 k ⎢ p ⎜ − tt ⎟ ⎥ ⎣ ⎣⎢ ⎝ p ⎠⎦⎥⎦
z 1 where η = and ierfc()η exp( 2 )−−= erfc()ηηη 2 αt π
The properties of the wafer and the laser parameters used for calculating the temperature
distribution are listed in Tables 2.1 and 2.2 respectively. Figure 2.3 shows the transient
temperature distributions along the depth up to 1.3 μm for a n-type 6H-SiC wafer at time
t = 64 ns and for a p-type 4H-SiC wafer at time t = 89 ns, which represent the
temperature fields during laser irradiation since the pulse-on times tp = 65 and 90 ns for the 6H- and 4H-SiC wafers respectively. During the laser pulse-off time, the transient temperature distributions along the depth up to 1.3 μm are given in Figure 2.4 at 5 and 1
64 ns immediately after the laser pulse was off, i.e., at times t = 66 and 91 ns after the laser irradiation began, for the 6H- and 4H-SiC wafers respectively.
The maximum surface temperatures reached by the 6H- and 4H-wafers are 2898
K and 3046 K respectively as shown in Figure. 2.3 These temperatures were selected for laser doping in the respective substrates because thermal damages, such as melting of the surface at isolated spots, were found to occur beyond these temperatures.
2800 4H-SiC (p-type) 6H-SiC (n-type)
2600
2400
Temperature (K) Temperature 2200
0 200 400 600 800 1000 1200 1400 Depth (nm)
Figure 2.4. Transient temperature distributions along the depths of n-type 6H-SiC and p- type 4H-SiC wafers at times t = 70 and 91 ns respectively after the laser irradiation began.
65 2.3 Laser Doping Experiment
2.3.1 Sample Preparation
A typical cleaning procedure carried out on SiC wafer substrates prior to laser doping:
1. Hold under running tap water and then rinse it with de-ionized (D.I) water
2. Scrub with soap and then rinse with D.I water
3. Dipping in (H2SO4 and H2O2) (1:1) for 15 min
4. Rinse it with D.I water
5. Buffered Oxide Etch (BOE) etching to etch out any native oxide of Si formed on the surface for 10 mins. Use a plastic watch glass instead of glass as BOE etched glass.
6. Rinse with D.I water
7. Clean with acetone followed by methanol followed by rinsing with D.I water
8. Blow dry with inert gas or nitrogen.
2.3.2 Experimental Setup
A schematic of the laser doping system is shown in Figure 2.5. A clean sample is placed inside a doping chamber and the chamber is pumped down to 1 mTorr or lower vacuum using a combination of mechanical pump and a diffusion pump. A stainless steel bubbler immersed in the water bath delivers the dopant gas to the doping chamber along with a carrier gas. Sample is simultaneously irradiated with the laser and exposed to a dopant- containing ambient. A programmed (Appendix A) X-Y stage from Velmex Inc. guides the laser to the desired area of doping.
66
Figure 2.5. Experimental setup for laser doping of silicon carbide substrates for p-n junction and LED fabrication
2.3.3 N-type doping (N, Se)
For n-type doping LISPD (dopant gas method) was used. For nitrogen, highly pure nitrogen gas was used as a dopant source. While for selenium, an organometallic compound of selenium (diethyl Selenide) was used as a dopant source. Se precursor was heated in a bubbler immersed in a water bath maintained at 100°C. When the precursor evaporated a carrier gas, argon, was passed through the bubbler to transport the vapor to the laser doping chamber. The precursor decomposes at the laser-heated surface producing Se atoms which subsequently diffuse into the wafer. The sample was placed in dopant atmosphere at pressure of 30 psi. Laser-doped tracks were formed on the sample surface by moving the chamber with a stepper motor-controlled translation stage which is
67 pre-programmed to dope the desired locations on the wafer surface. The height of the chamber was controlled manually through an intermediate stage to obtain different laser spot sizes on the SiC substrate surface.
2.3.4 P-type doping (Al, Cr and B)
For p-type doping, the precursors were: (1) trimethylaluminum (TMA) and Bis
(ethyl benzene)-chromium for LISPD (dopant gas method) (2) Boron powder (~325 mesh, 99.9% purity) for LISPD (dopant film method).
In case of LISPD, the organometallic precursor was heated in a bubbler immersed in a water bath maintained at 70-100°C depending on its boiling point. When the precursor evaporated a carrier gas, argon, was passed through the bubbler to transport the vapor to the laser doping chamber. The precursor decomposes at the laser-heated surface producing respective atoms which subsequently diffuse into the wafer.
In the case of boron doping, dopant film method was used in which a bed of boron powder was sandwiched between the SiC wafer and a soda lime glass slide at the wafer surface. The soda lime glass allows the incident Nd:YAG laser (1064 nm) beam to reach the SiC wafer surface, restrains the boron powder at the SiC surface and confines the molten precursors to the surface. In this study, the dopant film method was a two-step process: the Nd:YAG laser was first run to melt the boron powder bed in order to create a solid film of B on the surface of SiC wafer and then the laser was used to heat the solid boron film in order to diffuse B into the SiC lattice. The SiC substrate was kept at room temperature before laser irradiation in all the experiments. The laser-processing parameters are listed in Table 2.4.
68 Table 2.4. Nd:YAG (1064 nm) laser doping process parameters for SiC LED fabrication.
Sample Color Dopant Power Pulse Focal Spot # of Scanning Dopant Contribution (W) repetition Length size passes speed medium in LEDs rate (mm) (μm) (mm/sec) (KHz)
4H:SiC White, Blue, N 12 5 150 80 3 0.8 Ultra high pure Voilet nitrogen 30psi p-type White, Cr 12.5- 5 150 65 1 0.5 Bis (ethyl Orange-Red 13 benzene)- chromium, argon 30 psi
Blue, Red, Se 11.5- 5 150 100 2 0.8 Diethyl White 13 selenium, argon 30 psi
6H:SiC White, Cr 11.6 4 150 65 1 0.8 Bis (ethyl Green benzene)- n-type chromium, argon 30 psi
White, Blue Al 11.5- 5 150 65 2 0.5 Trimethyl 12 aluminum, Methane 30psi
White, B 10.5 CW 150 100 1 0.8 Boron Powder, Blue- Green argon 30 psi
White, B 12 5 150 80 2 0.8 Drive in, Blue-Green Argon 30 psi
4H:SiC White, Red, Cr 12 5 150 65 2 0.5 Bis (ethyl Green, benzene)- n-type Voilet chromium, argon 30 psi
All samples were post-cleaned with a 45% by wt. KOH solution and then rinsed
with acetone, methanol and D.I. water before depositing the contacts for electrical and
optical characterizations. Al, Cr, Ni, Au, Cu and In films were used and studied as a
contact metal for these SiC devices (Appendix C). A film of ~250-500 nm thick was
thermally evaporated to form a p and n side contacts. Thermal annealing in Argon
69 environment for 15-20 mins. at 450°C was also performed on certain devices post contact deposition to improve the contact properties. Laser doping studies (Appendix B) were also performed on other material systems such as GaP, Si, glass and undoped SiC using dopants such as Pd, P, Al, B and N.
70 CHAPTER 3: CRYSTALLINE QUALITY, ELECTRONIC AND ELECTRICAL PROPERTIES ANALYSIS
3.1 Crystalline quality analysis
3.1.1 Effect of laser doping on crystalline quality
Doping of SiC with conventional or unconventional dopants using ion implantation or thermal doping poses severe drawbacks. [Harris (1995)]. Ion implantation suffers several difficulties such as (i) Incongruent evaporation of Si from the
SiC wafer surface during post implantation annealing which limits the maximum annealing temperature that can be used to repair the lattice damage or cause dopant activation, (ii) Restoration of the lattice quality back to the original level if the as- implanted lattice damage is near the amorphous level, (iii) Polytypic transformations may occur during post implantation annealing and (iv) Stoichiometric imbalances can occur leading to Si or C islands. Thermal doping is associated with the problems of DASE
(Damage-Assisted Sublimation Etching) of Si and dopant (e.g. Al or N) out-diffusion. In situ epilayer doping is another technique which alters the planarity of the epilayer, reduces the wafer yield and the doping process becomes too complex for producing different dopant profiles for different devices on the same wafer [Harris (1995), Bäuerle
D. (2002), Tian (2006)].
Laser-induced solid phase doping [Sengupta (2001), Salama (2003), Tian (2006)] technique is used to dope SiC with Cr. It involves no melting of the wafer and relies on solid state diffusion for dopant incorporation. Tian et al. (2006) incorporated carbon atoms into the SiC lattice and showed almost no crystalline damage. The degree of damage to the SiC surface and lattice during laser doping of Cr in SiC are analyzed in
71 this study.
3.1.1.1 Surface roughness and chemistry analysis using optical profilometer and EDS
Melting and ablation of the surface, and agglomeration of tiny dopant particulates on the surface are possible mechanisms of surface damage during laser doping, which strongly affect the surface roughness and chemistry. The surfaces of the original wafer and the laser-doped region were scanned with a Veeco (DEKTAK 3) optical interferometric profilometer equipped with a white light source and a red filter of wavelength 650 nm and an objective of 10X magnification. The laser doping process did not damage the substrate surface; instead the smoothness of the surface was improved as shown in Figure 3.1(a), indicating that the average surface roughness decreased from
1.92 nm of the original wafer to 1.69 nm after the laser doping process.
The chemistry of the surface was analyzed using energy dispersive X-ray spectrometry (EDS) by scanning a cross-section of the laser-doped region along the line
ABCD as shown in Figure. 3.1(b). The EDS data are presented in Figure. 3.1(c) showing the concentrations of Si and C from points A to D for a total distance of 130 nm. The line segment AB is in the Au-Pd eutectic region where Si is absent but a certain amount of C is present. The line segment BC is in the amorphous SiC region, which is formed during
FIB milling, where the concentrations of C and Si are found to increase gradually. No Si is detected in the top 17 nm thick layer (Figure 3.2) of this amorphous region, which indicates that the Si atoms of this layer might have ablated or sublimed during FIB milling. These variations in compositions can also be attributed to other factors: (i) the
STEM mode, which was used for EDS analysis, utilizes a nanoprobe of size 2-5 nm and
(ii) the vibrations of the instrument and surroundings can cause sample movement. After
72 this amorphous region, the original SiC crystalline region is marked by the line segment
CD. The counts for Si and C are same in this region, indicating that no ablation or sublimation of the surface has occurred during laser doping.
35
d (b) Crystalline SiC (a) P (c) \ Crystalline 30 C D
Au B 25
SiC Amorphous SiC AB C D 20 15
10
SiC C Au/Pd Eutectic 5 Si Counts (arbitrary units) 50 μm 20 nm Amorphous A 0 0 20 40 60 80 100 120 140 Position (nm)
Figure 3.1. (a) Optical interferometric micrograph of laser Cr-doped 4H-SiC (p-type) wafer surface showing no surface damage and improved surface roughness after laser doping. (b) TEM micrograph showing the cross-section of the sample and the path for
EDS line scan. (c) EDS scan starting from A in the Au-Pd thin film region to D in the crystalline SiC region, showing the chemistry of SiC wafer.
3.1.1.2 Crystal lattice analysis using FIB and TEM
To examine the crystalline quality of the laser-doped sample, cross-sections of laser Cr-doped 4H-SiC wafer was prepared by Focused Ion Beam (FIB) milling in FEI
Nova 600 system equipped with an omni-probe for Transmission Electron Microscopy
(TEM) sample preparation capability. TEM studies were conducted using the JEOL 2010 system operated at 200 kV. Tian et al. (2006) investigated crystalline damage in n-type
6H-SiC owing to laser incorporation of Al, N and C using TEM and Rutherford
73 backscattering (RBS) to detect any amorphization of the wafer. These atoms are close to the size of one or the other host atoms in SiC. No studies have been reported so far on defect generation due to the incorporation of large diameter dopants such as Cr into SiC.
Okojie et al. (2002) have shown that there is a polytypic transformation from 4H-SiC to
3C-SiC during dry oxidation annealing at 1150°C. Therefore it is important to investigate any changes that may occur in SiC during laser doping.
TEM micrographs of laser Cr-doped 4H-SiC wafer are presented in Figure 3.2 to show that the laser doping process does not generate any extended defects in the wafer.
Figure 3.2a shows two distinct amorphous layers, a 3 nm thick layer at the top surface of the laser-doped specimen followed by the second layer of thickness 13 nm. Tian et al.
(2006) reported amorphous layers of thickness ~17 nm on laser-doped 6H-SiC wafers as well as on parent wafers and, therefore, attributed the formation of amorphous layers during platinum deposition while preparing the TEM samples. The platinum film was deposited by Ga+ ion beam-assisted CVD (chemical vapor deposition) using an accelerating voltage of 30 keV with 2.2 μA beam current. Even 40 nm thick sputter- deposited Au film on the laser-doped and parent wafers before FIB operation could not prevent the substrate damage during the subsequent platinum film deposition process.
They concluded that the gold films were not thick enough to prevent amorphization at the sample surface.
In the present work, a 70 nm thick Au-Pd film was deposited prior to Pt film deposition to reduce the effect of amorphization. Then a 180 nm thick Pt was deposited over the Au-Pd layer to prevent FIB-induced damage as shown in Figure 3.2a. A distinct
3 nm amorphous layer appears to be different from the extended 13 nm amorphous layer.
74 However, the entire amorphous region of 16 nm might have formed due to polishing while fabricating the parent wafer [Nakashima et. al. (2006)] or due to the sputter- deposition of Au-Pd film in a vacuum chamber [Oswald et. al. (1999)] and Ga+ ion beam- assisted CVD of Pt inside the FIB milling system [Tian (2006), Kempshall (2002)].
Another possibility could be that the entire 16 nm
Platinum Amorphous SiC (a) (b) (c)
Amorphous SiC Gold-Palladium Crystalline 4H-SiC(p-type)
Amorphous SiC
4H-SiC(p-type)
20 nm 5 nm Crystalline 4H-SiC(p-type) 5 nm
Figure 3.2 (a) TEM micrograph of the laser Cr doped 4H-SiC substrate. (b) 16 nm thick amorphous layer, Au-Pd film and SiC lattice. (c) High resolution image is SiC lattice extending from the amorphous region in (b) into the SiC lattice and TEM diffraction pattern indicating a single crystal SiC pattern with no signs of amorphization or defect generation post laser doping.
amorphous layer was formed during the Pt film deposition on top of the Au-Pd film deposition because the fusion temperature of Pt, 2041 K, is much higher than the eutectic temperature, 373 K [Okamoto (1985)], of Au-Pd at the composition of 37.5 wt.% Au. So the amorphous layer may not form during the Au-Pd film deposition due to relatively low speed of the sputtered Au-Pd particles and low deposition temperature at the Au-Pd
75 eutectic composition. In the case of Pt film deposition, however, the speed of the Ga+ ion beam-assisted CVD Pt particles as well as the film deposition temperature would be higher, which can heat up a thin layer of SiC at the SiC-film interface. Therefore, the particles with high kinetic energy are expected to disorder the crystal lattice. When the Pt film deposition ends, the disordered layer cools down and a very thin layer adjacent to the metal layer can cool down faster by heat conduction to the metal layer than the rest of the disordered layer. This very thin layer (3 nm in this study) manifests itself as a distinct amorphous layer compared to the rest of the disordered layer. A high magnification
TEM image shows the crystalline integrity beneath the amorphous layer in Figure 3.2b, which is identical to the crystalline lattice of the parent wafer observed deep inside beyond the laser-doped region. Stacking faults, lattice disorders, dislocations, slips, low- angle grain boundaries, point defects (such as vacancies and interstitials) and other polytypic transformations are commonly observed in SiC after high temperature annealing following ion implantation or thermal doping [Harris (1995), Tian (2006),
Okojie et al.(2002)]. No such defects were observed in the laser-doped sample as evident from the TEM micrograph. Figure 3.2c shows TEM diffraction pattern of the laser-doped sample, which corresponds to a single crystal 4H-SiC and further, validates the crystalline integrity of the sample. Cr atoms or any other dopant atoms (e.g., Al atoms that pre-existed in the sample in this study), however, are bound to cause some lattice distortion or strain in the SiC lattice due to the difference in the atomic radii of the host and dopant atoms. It, therefore, can be concluded that the laser doping process causes less damage to the parent wafer than other conventional doping techniques such as ion- implantation and thermal doping.
76 3.1.2 SIMS studies for dopant concentration and solid solubility analysis
Laser doping readily facilitates the formation of dopant concentration gradients along the depth of the wafer, i.e., the concentration decreases gradually along the depth.
Thus laser-doped samples are inherently adaptable to linearly graded junction fabrication, which can be useful for various semiconductor device applications [Berch et. al. (1976)].
It is easy to fabricate ultra shallow or ultra deep junctions using laser doping. Since the laser beam is rastered over the surface during doping, it becomes matter of concern to know the uniformity of dopant distribution over that entire area. This study has been performed in Si by laser doping it with Al. The SIMS surface scans were obtained at surface and depths of 150 nm and 300 nm. The doping was very uniform over the entire region. [Appendix B: Laser doping studies]
SIMS analysis was carried to obtain the dopant concentration profile in the SiC wafer substrate post laser doping. Both white LED samples, laser Cr 6H-SiC (n-type) and
Cr doped 4H-SiC (p-type) samples were analyzed. A Cr-implanted SiC standard (1×1020 cm-3), aluminum-doped SiC standard (1x1019 cm-3), N-doped SiC standard (5x1018 cm-3) and two as-received SiC samples, n-type (5×1018 cm-3) 6H-SiC and p-type (1×1019 cm-3)
4H-SiC, were also analyzed for reference and background concentrations respectively.
3.1.2.1 Enhancement of solid solubility for all the dopants
Figure 3.3 indicates that the maximum concentration of Cr in 4H-SiC and 6H-SiC are 1.42×1019 cm-3 and 2.29×1019 cm-3, respectively, which are almost two orders of magnitude higher than the reported solid solubility limit (3×1017 cm-3) for Cr [Harris
(1995)]. Similarly enhanced solid solubilities have been observed for other dopants (Al
77 ) 6H-SiC (n-type) -3 1019 4H-SiC (p-type) background
1018
1017
1016 Cr Concentration (cm Cr Concentration
1015
0 40 80 500 1000 1500 Depth (nm)
Figure 3.3. SIMS analysis for the concentration profiles of laser doped Cr, along the depth of the 6H-SiC (n-type) and 4H-SiC (p-type) substrate. The penetration depth is 1.5
μm in the case of 6H-SiC and 80 nm in the case of 4H-SiC.
Table 3.1. Increase in the concentration beyond the solid solubility limit of various dopants in silicon carbide by laser doping.
Element Cr Al N B
Literature 3×1017 2×1021 6×1020 2.5×1020 [cm-3] Laser doping 2.29×1019 2.5×1022 2×1021 5×1020 [cm-3]
and N) as well [Tian (2006)]. Table 3.1 shows the comparison of the solid solubilities for various dopants in SiC.
78 3.1.2.2 Diffusion of large sized atom (Cr) in SiC
Non-equilibrium heat and mass transfer processes enable dopant incorporation exceeding the solubility limit. During the laser doping process, the wafer is under non- isothermal condition due to localized heating. The wafer temperature is raised to sufficiently high temperatures easily in a very short time. The rise in temperature is non- uniform along the thickness as well as the lateral direction, leading to the lattice expansions by different amounts in different directions. Thus thermal stresses are generated between the heated region and the surrounding cooler region promoting stress- induced diffusion. Since the diffusion is much less in the cooler region than in the hotter region and the temperature varies predominantly along the thickness of the wafer compared to the lateral direction, the laser doping process promotes the diffusion of atoms along the direction of the laser beam propagation. The vacancy concentration can be much higher than their equilibrium values due to higher temperature, thermal diffusion, thermal expansion, laser-induced shock waves, and electronic and vibrational excitations [Bäuerle (2002)]. Additionally the rapid heating and cooling due to scanning pulsed laser beam activates a locking mechanism especially for large sized atoms such as
Cr thus facilitating greater non-equilibrium mass flux into the substrate.
From the SIMS profiles in Figure 3.3, it is evident that the penetration depth of Cr is much higher in 6H-SiC than in 4H-SiC, which can be explained on the basis of the types and concentrations of dopants in the as-received wafers, and the laser doping process. It can be observed in Table 3.4 that higher fluence was used for doping 6H-SiC than for 4H-SiC. So the above-mentioned thermal effect on diffusion is more for 6H-SiC than for 4H-SiC, leading to deeper dopant profile in the case of 6H-SiC.
79 In the case of 4H-SiC, aluminum is the prominent dopant with concentration 1019 cm-3. The atomic radius of aluminum is 0.118 nm which is very close to the atomic radius of silicon 0.11 nm, while the atomic radius of Cr is 0.14 nm which is greater than that of Si or C (0.07 nm). Therefore Cr is more likely to occupy a Si vacancy than a C vacancy. Since most of the Si vacancies are already occupied by aluminum atoms in 4H-
SiC, additional vacancies generated during laser doping leads to competition between Al and Cr atoms for vacant site occupancy. The concentration of Al already being much higher than that of Cr, the aluminum atoms are expected to inhibit the diffusion of Cr.
Additionally the diffusion of Cr has to be accompanied by the lattice strain due its larger radius.
In the case of 6H-SiC, however, nitrogen is the prominent dopant with concentration 5×1018 cm-3. Nitrogen (atomic radius 0.065 nm), which is a smaller atom than Si or C, can occupy either a C or Si site with a higher probability of occupying a C vacancy due to similarity in the sizes of C and N. Therefore Cr atoms can occupy all of the Si vacancies that are already present in the as-received wafer and generated during laser doping. For these reasons, the dopant penetration depth and the surface concentration of Cr are higher in the n-type 6H-SiC than in the p-type 4H-SiC.
3.1.2.3 Diffusion of Al, Cr and N for SiC White LEDs
The same Cr doped samples were further doped with Al and N respectively to fabricated White LEDs. Figures 3.4 and 3.5 show the SIMS spectrometry results for both the 6H and 4H-SiC white LED samples. Figure 3.4 shows that a very high concentration
80 of about 2x1020 cm-3 was measured for Al at the surface decreasing gradually to 1x1018 cm-3 to a depth of 700 nm, while a concentration of 1x1019 cm-3 was measured for Cr at ) -3 White LED 6H-SiC (n-type) 1E20 Cr-Std
Al-Std
1E19 Al
6H-SiC (N) 1E18
Cr Undoped
Cr, Al, N Concentration (cm Cr, Al, N Concentration 1E17 0 100 200 300 400 500 600 700 Depth (nm)
Figure 3.4. SIMS analysis of white LED fabrication on 6H-SiC (n-type N) substrate laser doped with Cr and Al showing the variation of concentration along the depth of the substrate.
the surface decreasing gradually to the background concentration of 1x1017 cm-3 at a depth of 420 nm for 6H-SiC (n-type) white LED sample. The experimental parameters,
Table 3.4, used for doping Al and Cr are similar except for a small variation in the pulse repetition rate. Salama et. al. (2006) showed that the penetration depth increases with the increasing pulse repetition rate, however, differences between the rates were two of orders of magnitude; in this study it is an insignificant effect. The penetration depth of Al is much larger than that of Cr due to the size effect of these elements. Aluminum is much smaller than Cr and slightly smaller than Si and it can penetrate SiC easily and occupy the available Si vacancies. For Cr occupancy of Si vacancies the penetration is
81 accompanied by lattice strain. Similarly, Figure 3.5 shows that a very high concentration of approximately 1x1021 cm-3 was obtained for N at the surface decreasing gradually to
3x1017 cm-3 to a depth of 8.5 μm, while a concentration of 1.5x1019 cm-3 was obtained for
Cr at the surface decreasing gradually to the background concentration of 1x1015 cm-3 at a depth of 80 nm 4H-SiC (p-type) white LED sample. Essentially the same parameters were used for doping both the elements except for a small variation in the pulse repetition rate. The penetration depth of N is much larger than that of Cr, which results from the atomic radius difference. The smaller nitrogen atom can penetrate SiC easily and occupy the available C or Si vacancies.
1E21 1E21 ) -3 Al-Std White LED 4H-SiC (p-type)
1E20 )
Cr-Std -3 Al 1E19 1E20
1E18 N-Std N
1E17 Cr 1E19 As received substrate 1E16 Cr, Al Concentration (cm N Concentration (cm N Concentration 1E15 1E18 0 100 7000 8000 Depth (nm)
Figure 3.5. SIMS analysis of white LED fabrication on 4H-SiC (p-type Al) substrate laser doped with Cr and N showing the variation of concentration along the depth of the substrate.
82 3.2 Electronic Properties Characterization
3.2.1 Analysis of Cr energy levels and electronic defect states using DLTS
3d transition metal elements such as titanium, vanadium and chromium are the most commonly identified impurities in SiC. These elements usually act as compensating sites which are useful in producing semi-insulating materials [Harris (1995), Justo
(2006)]. Intentionally introduced impurities or dopants are important in optimizing the electrical and photonic properties of SiC devices. The transition metal elements with the open d shell can have several different charge states corresponding to different electronic configurations, which in many cases give rise to deep level states in the band gap that are electrically and optically active [Lebedev (1999)]. These deep levels may act as efficient recombination centers, which limit carrier lifetimes or compensate for shallow dopants.
The behavior of Cr in SiC as a donor or an acceptor impurity is not clear.
Hemstreet (1977) proposed the following mechanism for bonding of Cr with Si atoms.
Neutral chromium has six valence electrons. Four of these electrons become involved in the formation of tetrahedral bonds with the neighboring silicon atoms, repairing the broken bonds introduced by the creation of a silicon vacancy, while the remaining two electrons go into a nonbonding atomic d-like impurity levels in deep Si band gap. Thus chromium behaves as a donor with calculated energy level of Ev+0.69 eV where Ev is the valence band edge. Achtziger et. al. (1997) experimentally determined the levels in Cr- implanted 4H- and 6H-SiC by deep level transient spectroscopy (DLTS). They concluded that chromium behaves as an acceptor with a deep energy level of Ec-0.54 eV for 6H-SiC and Ec-0.74 eV for 4H-SiC and a shallower energy level of Ec-0.15 eV and Ec-0.17 eV in
4H:SiC. In contrast Justo et. al. (2006) recently showed that chromium behaves as a
83 donor as well as an acceptor depending on its charge states. We present DLTS results in this work showing Cr as an acceptor.
The importance of Cr as a dopant is that it can act as a double acceptor as shown in this study, i.e., a single Cr atom can produce two holes. This can be understood based on its electronic configuration 1s2 2s2 2p6 3s2 3p6 4s1 3d5 which has two half-filled outer shells. Cr (atomic radius 0.14 nm) may occupy the silicon (atomic radius 0.11 nm) or carbon (atomic radius 0.07 nm) lattice site when doped into SiC. However the tendency to occupy the Si lattice site is high due to the size effect. In either case of site occupancy the silicon or carbon atom would share its 4 electrons with the 4s and 3d shells of each chromium atom creating two holes per Cr atom. Thus chromium acts as a double acceptor impurity, whereas conventional p-type dopants, such as Al, provide a single acceptor site per dopant atom.
The additional hole created by Cr may be useful in tailoring the luminescence properties of SiC LEDs based on the Donor-Acceptor Pair (DAP) recombination mechanism [Harris (1995)]. Additionally fewer Cr atoms in the SiC lattice are expected to produce less strain and correspondingly fewer defects in the substrate compared to higher concentrations of single acceptor for the same hole concentration. Cr in its inactive state can also provide red luminescence by the atomic transition mechanism similar to those observed in the case of rare earth elements [Denbaars et. al. (2006)].
Complete activation of the incorporated Cr is an additional concern. This work investigates the energy levels and electronic defect states formed by Cr in SiC and studies the activation of Cr atoms within SiC substrate post laser doping.
A Deep Level Transient Spectroscopy (DLTS) system developed by Sula
84 Technologies was used for determining the energy levels of chromium in SiC and to identify any additional electronic defect levels generated due to laser doping. This technique [Lang (1974)] involves a fast thermal scan of all the defect levels that are electrically active in the junction region of the diode. The following important information about the defects can be extracted from the DLTS spectra; (i) the saturated peak height is directly proportional to the density of defect level, (ii) the temperature at which the peak occurs is related to the ionization energy of the defect level, (iii) the sign of the peak, i.e., a positive or a negative peak indicates whether it is due to minority or majority carriers emission respectively and (iv) the carrier capture cross-section can be determined directly from the dependence of the DLTS peak height on the input electrical pulse (forward bias) duration. The temperature of the diode is varied by heating it with a thermal stage and simultaneously an electrical pulse is applied to the diode to generate a non-equilibrium condition due to the charge carriers diffusing to the depletion region during the pulse-on time. Some of these charges may be trapped in the depletion region depending on the temperature of the diode. When the pulse is off, the trapped charges are released and they try to diffuse out of the depletion region. The transient capacitance change in the depletion region associated with the return to the thermal equilibrium by the charge carriers from the non-equilibrium condition was measured during the DLTS experiment.
To carry out DLTS tests, a p-n junction was fabricated using the n-type (N-doped)
6H-SiC wafer that was laser-doped with Cr and Al. For p-contact and n-contact, an Al and a Ni film of approximately 250 nm thick each were thermally evaporated on the p- region (Cr-Al doped) and n-region (N-substrate) respectively. Similarly for the p-type (Al
85 doped) 4H-SiC wafer, Cr was doped on one side and N on the opposing side of the wafer and then. p- and n-contact metal films were fabricated as in the case of 6H-SiC sample.
Since the leakage currents in these devices have to be minimal for carrying out DLTS measurements, the contacts were annealed for 10 minutes at 350°C to obtain good Ohmic contacts.
The diode was heated from 300 - 400 K while maintaining a reverse bias (0.5 V).
The depletion region is wide and the junction capacitance is small under the reverse bias condition. A forward bias of 0.5 V was then applied to the diode as a pulse at 1 MHz to inject minority carriers (holes) into the depletion region to fill up the available hole traps.
This increases the positive charge density on the n-side of the depletion region and thus increases the junction capacitance by ΔC(0). After the forward bias pulse ends, the trapped holes are thermally re-emitted to the valence band causing time-varying p-n junction capacitance, which may be approximated as [Lang (1974)]:
⎛ t ⎞ C() ΔΔ exp)0(Ct ⎜−⋅= ⎟ (3.1) ⎝ τ ⎠ where τ is the trapped hole emission time constant. Since the temperature of the junction varies with time, the DLTS scan at two different temperatures, S(τ), is equivalent to taking the difference of Eq. 3.1 at two times t1 and t2, i.e.,
tt ⎛ 1 −− 2 ⎞ τ )0()()( ⎜ −⋅Δ=Δ−Δ= eeCtCtCS ττ ⎟ (3.2) () 1 2 ⎜ ⎟ ⎝ ⎠
The maximum emission rate, 1/τmax, can be obtained by setting dS(τ)/dτ = 0, which yields
− tt 21 τ max = (3.3) ⎛ t1 ⎞ ln⎜ ⎟ ⎝ t2 ⎠
86 It should be noted that the DLTS signal, S(τ), reaches its maximum at τmax corresponding to the characteristic temperature Tm. During the DLTS experiment, τmax is varied by changing the delay time variable, and a set of characteristic temperatures (Tm) is obtained for different delay times (~τmax). Heating and cooling cycles at 6°C/min ramp were employed during the DLTS measurements. The changes in the junction capacitance
(DLTS signals) were measured for different temperatures ranging from 300 to 400 K for both of the heating and cooling cycles and the delay time was varied from 0.02 to 0.2 ms.
Generally a minority carrier trap produces a positive DLTS peak, while a majority carrier trap yields a negative DLTS peak.
Positive DLTS peaks were observed for Cr-doped 6H-SiC and 4H-SiC as shown
20 0.02 ms -2.5 0.05 ms E + 0.458 eV v 15 0.1 ms -3.0 0.2 ms
10 )
n -3.5 Y = A + B * X /E 5 2 -4.0 Parameter Value Error DC (pF) DC ln(T ------0 A 10.5 7.9 -4.5 B -5.3 2.9 -5 ------(a) (b) -5.0 300 320 340 360 380 400 2.0 2.2 2.4 2.6 2.8 3.0 -1 Temperature (K) 1000/T (K )
Figure 3.6 (a) DLTS spectrum for 6H-SiC (n-type) substrate doped with Cr. A positive
DLTS signal was observed indicating that Cr is more of an acceptor impurity with holes
(minority carriers) as traps. (b) Activation energy calculation from the DLTS signals in
(a) for different delay times showing that Cr forms an acceptor level of Ev + 0.458 eV in
6H-SiC.
87 in Figures 3.6a and 3.7a indicating that Cr is an acceptor impurity with holes (minority carriers) acting as traps. Since the activation energy can be estimated from the Arrhenius plot given by
⎛ e ⎞ ⎛ 1 ⎞ ΔE ⎛ 1 ⎞ ⎜ p ⎟ = lnln ⎜ ⎟ ln A σ −⋅= ⋅⎜ ⎟ , (3.4) ⎜ 2 ⎟ ⎜ 2 ⎟ ()h ⎜ ⎟ ⎝ Tm ⎠ ⎝τ max ⋅Tm ⎠ ⎝ Tk m ⎠ the DLTS signals were plotted in Figures 3.6b and 3.7b to obtain linear variation in the
⎛ e ⎞ data with Y = ln⎜ p ⎟ and X = 1/T . The activation energy determined from the slope ⎜ 2 ⎟ m ⎝ Tm ⎠ of the straight line shows that the Cr dopant atoms form an acceptor level of Ev + 0.458 eV in 6H-SiC (Figure 3.6b) and Ev + 0. 8 eV in 4H-SiC (Figure 3.7b). These energy levels are different from the reported acceptor levels for Cr, i.e., Ev + 0.54 eV in 6H-SiC and Ev + 0.74 eV in 4H-SiC [Harris (1995),
1.0 -2 0.8 0.1ms E +0.8 eV 0.2ms v -3 0.6 )
n Y = A + B * X 0.4
/E -4 2 Parameter Value Error 0.2
ln(T ------DC (pF) DC -5 0.0 A 25.5 12.14 B -9.3 3.8 -0.2 -6 ------
300 325 350 375 400 2.4 2.7 3.0 3.3 -1 Temperature (K) 1000/T K
Figure 3.7 (a) DLTS spectrum for 4H-SiC (p-type) substrate doped with Cr. A positive
DLTS signal was observed indicating that Cr is more of an acceptor impurity with holes
(minority carriers) as traps. (b) Activation energy calculation from the DLTS signals in
(a) for different delay times showing that Cr forms an acceptor level of Ev + 0.8 eV in
4H-SiC.
88 Lebedev (1999) and Achtziger et. al. (1997)]. This discrepancy may be due to several factors such as the parent wafer quality, laser doping technique, contacts, leakage currents, and the operational characteristics of the DLTS equipment. However, these results confirm that Cr is an acceptor impurity, supporting the determination of the impurity state based on the electronic configuration (1s2 2s2 2p6 3s2 3p6 4s1 3d5) of Cr and the sharing of its electrons with either Si or C atoms as explained earlier. No additional electronic defect states were observed besides Cr levels, confirming that laser doping does not generate any electronic defects in SiC due to Cr doping.
3.2.2 Analysis of activated state of dopants using Hall Effect measurement
The active state of laser-doped Cr in n-type (nitrogen-doped) 4H-SiC was determined using an Ecopia Hall effect measurement system at room temperature. This sample was prepared using the same doping parameters as for doping Cr into the p-type
4H-SiC wafer, and SIMS analysis was carried out to obtain the Cr concentration profile in the sample. The Hall measurement identified the Cr-doped region as p-type with a carrier concentration of 1.942×1019 cm-3 which is almost twice the average dopant concentration (~1019 cm-3). The average dopant concentration is based on the SIMS data with a concentration of 2×1019 cm-3 at the wafer surface and 1017 cm-3 at a depth of 500 nm. These data confirm Cr as a double acceptor and also indicate that the laser doping technique itself produced almost all of the dopants in the activated state without requiring any annealing step for dopant activation.
89 3.3 Electrical Characterization
3.3.1 P-N junction device fabrication for study of electrical properties
Four p-n diodes were fabricated separately to study the I-V characteristics and dopant concentration effect. Figure 3.8a shows the geometry of three of the diodes (N1,
N2, N3) for which an n-type 6H-SiC wafer was doped with aluminum to form a p-region and the aluminum and indium contacts were made at the top p region and the bottom n region respectively. For the fourth diode (blue LED sample), an additional n+ region was created by laser doping the wafer with nitrogen on the top surface beside the p-region for establishing Ohmic contact with reduced contact resistance. The distance between the p- and n-doped regions was approximately 4 mm. Both the indium and aluminum contacts were made at the top surface for the blue LED sample as shown in Figure 3.8b, while the contacts were made at the top and bottom surfaces of the wafer for N1, N2 and N3 samples to investigate the effects of different contact configurations.
Figure 3.8 Laser-doped device geometry: (a) p-n diode with laser-doped p (Al) region and as- received n region of the n-type 6H-SiC substrate and (b) p-n diode with laser- doped p (Al) region and as-received n region, and an additional laser-doped n (N) region for improved contact and LED performance.
90 The experimental parameters used for doping of these samples are listed in Table 3.2 and
Figure 3.9 show the SIMS concentration profiles for these samples.
Table 3.2 Experimental parameters for laser doping of N1, N2, N3 and blue LED.
Sample Dopant Power Pulse Focal Spot # of Scanning Dopant medium 6H:SiC (W) repetition Length size passes speed rate (mm) (μm) (mm/sec) n-type (KHz) N1 Al 11.5-12 3 150 65 2 0.5 TMAL, Ar 30psi N2 Al 12.5-13 5 150 65 2 0.5 TMAL, Ar 30psi N3 Al 12.5-13 10 150 65 2 0.5 TMAL, Ar 30psi Blue N 13 10 150 100 1 0.8 N 30psi LED Al 10.3 3 150 100 3 1 TMAL, Ar 30psi
1E22 Standard Undoped )
-3 1E21 N1 N3 1E20 N2 Blue LED 1E19
1E18
Al Concentration (cm Al Concentration 1E17
0 50 100 150 200 250 300 Depth (nm)
Figure 3.9 SIMS analyses of p-n junction diodes and blue LED sample showing the variation of aluminum concentration with the depth of the substrate.
3.3.2 Measurement of C-V characteristics using LCR meter
C-V measurements were conducted to obtain the approximate charge carrier concentration in the substrate. A linear plot is generally expected for 1/C2 as a function of the bias voltage. However it is not a straight line as shown in Figure 3.10, indicating that
91 18 7.4x10-10 2.7x10
18 -10 2.6x10 7.2x10 12/ ⎡ ⎤ 18 ⎢ ⎥ 2.5x10 7.0x10-10 qε C = ⎢ s ⎥ 2.4x1018 ⎢ ⎛ 11⎞ ⎥ -10 18 6.8x10 ⎢ 2⎜ + ⎟−φ V ⎥ ) 2.3x10 ()ia -2 ⎣⎢ ⎝ NNad⎠ ⎦⎥
(F 18
2 2.2x10 6.6x10-10 18 1/C 2.1x10 -10 1 2(NNad+ ) 6.4x10 18 Capacitance (F) 2.0x10 2 = ()φia− V C qNNεsad -10 6.2x10 1.9x1018
18 -10 1.8x10 6.0x10 012345678 0123456789 Voltage (V) Voltage (V)
Figure 3.10 C-V characteristics of LED structure (Figure 3.8b) on 6H-SiC (n-type) substrate doped with aluminum
the dopant concentration varies with the depth of the sample. Since the temperature varies along the depth of the sample during laser doping, the diffusivity of the dopant is non- uniform along the depth and consequently a non-uniform dopant profile is formed. The flattening of the curve from 7 V onwards is due to series resistances involving contacts, depletion width and the n- and p- regions [Streetman (1990)]. The intersection of the 1/C2 curve with the horizontal axis yields a built-in voltage of 1.5-1.8 V and an effective
14 -3 acceptor concentration (Na) of 6.24x10 cm is obtained for the blue LED sample from the slope (0.108 nF-2V-1) of the curve for an approximate area of 32 mm2 using the following Eq.(3.5).[Streetman (1990)].
2 1 N a −= 2 ⋅ 2 (3.5) ε ⋅⋅ ()//1 dVCdAq aj
Where Na is the acceptor concentration, q is the electronic charge, ε is the dielectric
1 constant for SiC, 2 is the 1/slope and A is the area. As observed from the C- ()//1 dVCd aj
92 V measurements, the acceptor concentration is very low forming a p-n+ junction. Similar measurements were carried out to estimate the acceptor concentration in samples doped with varying p-type dopant concentrations where the contact area was about 6 mm2. Their charge carrier concentrations have been listed in Table 3.3.
3.3.3 Measurement of I-V Characteristics using I-V curve tracer
The I-V characteristics of these diodes and the as-received substrate were measured using a Tektronix 576 I-V curve tracer, which are presented in Figure 3.11 and
3.12. The additional laser-doped n+ region in the blue LED sample allows the formation of Ohmic contacts with reduced resistance. The contact area was 6 mm2 for all the samples. The I-V response is similar to the response of a typical p-n junction with a higher reverse breakdown voltage. These devices emitted blue light with their brightness increasing with increasing forward bias, indicating higher minority carrier (hole) injection at higher forward biases. Here the minority carrier injection is considered stronger than the majority carrier (electron) injection because the mobility of holes is less than the mobility of electrons in SiC [Harris (1995), Zhang et.al. (2001)].
3.3.3.1 Forward bias characteristics
The forward bias response of the p+pnn+ (blue LED) diode (Figs. 3.11a and 3.12) can be divided into three regimes based on the applied voltage: (i) non-response regime
(0-1 V) where the current is extremely small (nA) or almost zero until an appreciable amount of current begins to flow through the device after a specific threshold voltage, (ii) nonlinear response regime (1-8 V) where the current varies nonlinearly with the bias voltage and (iii) linear response regime (8-14 V) involving a sharp rise in the current.
93 Usually the limiting forward bias across a p-n junction is equal to the total potential due to the contacts, p- and n-doped regions and the depletion width. Both built-in potential across the junction and the total resistances (due to contacts and the p and n regions) across the device are operative below 1 V, and only the resistances and the tunneling of carriers affect the current flow across the device above 1 V. The nonlinear response after the threshold voltage is due to the low hole concentration gradient p region (p-drift region) and low hole mobility. The width of this region varies with dopant concentration.
With increase in the forward bias, the voltage drop keeps decreasing in the p-drift region due to the potential reduction caused by conductivity
As received Blue LED 125 500 N3 400 100 N2 300 75 N1 200 50
Current (mA) 100 25 0 0 -15 -10 -5 0 5 10 15 -15 -10 -5 0 5 10 15 -100 -25 Current (mA) Voltage (V) -200 -50 -300 -75 -400 -100 -500
-125 Voltage (V)) (a) (b)
Figure 3.11 I-V characteristics of different device structures: (a) as-received 6H-SiC (n- type) substrate with indium contacts, and p-n junction diode fabricated using the 6H-SiC
(n-type) substrate doped with aluminum; (b) comparison of I-V curves of three p-n junction diodes with different aluminum concentration profiles. modulation and self heating of the p-region; this leads to an increasing forward voltage drop across the junction region. Thus the injection efficiency of the junction continues to
94 increase in these devices with increase in the forward bias; this effect is referred to as the self adjusting emitting efficiency [Zhang et. al. (2001)]. The slope of the I-V curve, however, changes at a critical value of the bias voltage, indicating the onset of radiative recombination [Bergh et. al. (1976)]. The critical bias voltage is found to be 8 V in
Figure 3.12, which is also confirmed by the electroluminescence data. After this point, the linear response regime (8-14 V) occurs where the current rises sharply because the potential barrier of the device is overcome due to high injection and tunneling effects.
Also an increase in the intensity of the light emitted by the device was observed after 8 V, which is mainly due to high injection of holes as discussed earlier.
Similar trend was observed in the I-V characteristics of N1, N2 and N3 diodes as shown in Figure 3.11(b). Their nonlinear I-V response extends up to 15 V, which is higher than that of the p+pnn+ blue LED sample. This higher voltage may be due to the lack of an Ohmic contact at an n+ region and the device structure. To analyze this nonlinear response and the possible operative mechanisms for conduction, the average series resistances were calculated from their respective I-V curves for three different voltage ranges 0-8 V, 8-15 V and 0-15 V. The values of the resistances have been tabulated in Table 12 for different samples. RS1 is the average sheet resistance for the forward voltage range of 0-15 V, RS2 is the sheet resistance due to initial depletion region, p-drift region and tunneling through the depletion region in the voltage range 0-8
V and RS3 is the sheet resistance due to high injection and recombination in the voltage range 8-15 V.
95 10.0 30 9.5 Onset of 20 radiative 9.0 10
0 8.5 -18 -13 -8 -3 2 7 12 17
Current (mA) -10 8.0 Current (mA)
-20 7.5
-30 789101112 Voltage (V) Voltage (V) (a) (b)
Figure 3.12 Detailed I-V characteristics of a blue LED structure on 6H-SiC (n-type) substrate doped with aluminum showing the onset of radiative recombination in (a) and a magnified view of the onset region in (b).
Table 3.3 Series resistances and carrier concentrations measured using I-V and C-V respectively for the as-received substrate, 3 p+pn junction devices (N1, N2 and N3) and p+pnn+ blue LED .
Voltage As received Parameters N1 N2 N3 Blue LED range (V) substrate
RS1 (Ω) 0-15 145 171.7 162.9 98.08 94.2
RS2 (Ω) 0-8 2083 786 138.6 103.5 75.6
RS3 (Ω) 9-15 49.5 19 79 17.7 17.5
Avg. Carrier 17 17 17 16 14 3.4x10 2x10 3.5x10 4.5x10 6.24x10 concentration
96 The N1, N2 and N3 diodes showed decreasing trend in the calculated series resistances in all the three voltage ranges. This trend corresponds to the increasing laser pulse repetition rates used for doping the substrate for these diodes. As discussed earlier, the concentration and depth of the doped region vary with the pulse repetition rate and therefore affect the electrical properties of the p-drift region. The series resistances were
RS1= 94.2 Ω, RS2= 75.6 Ω and RS3=17.5 Ω for the LED sample, which are much lower than in the case of N1, N2 and N3 diodes. The lower resistances are due to n+ region below the indium contact and the planer (top surface) device structure.
The I-V characteristic of the as-received n-type 6H-SiC substrate was also analyzed by soldering indium contacts to the top and bottom surfaces of the wafer.
Indium forms mostly ohmic contact with n-type 6H-SiC wafer. The forward threshold bias was found to be 4 V as shown in Figure 3.12a, indicating that the built-in potential is
4 V. The built-in potential, however, is expected to be the same as the bandgap (3 eV) since the as-received sample was only an n-doped substrate. So the increase in the threshold voltage may be due to additional resistances at the indium contacts. The nonlinear I-V response regime is small, 4-8 V, compared to the diode cases. This is because high electron concentration in the n-type 6H-SiC as- received substrate facilitates the easy flow of electrons inducing the transition of nonlinear response to the linear response at the lower transition voltage of 8 V. Consequently a sharp linear rise in the forward current from 4 mA to 125 mA is observed after overcoming the potential across the device at 8 V. The transition voltage was higher for the above-mentioned p-n diodes due to the presence of the additional junction potential. This trend in the I-V characteristics indicates that diffusion and tunneling of electrons are dominant at low
97 voltages, while the electron injection and electron-hole recombination dominate at high voltages.
3.3.3.2 Reverse I-V characteristics
The reverse bias characteristic of the blue LED (Figures 3.11a and 3.12) shows a reverse breakdown voltage of 9 V with a soft preavalanche breakdown at 7-8 V with very low currents in the range of 0.01-0.4 mA. Soft preavalanche breakdown was also observed for N1, N2 and N3 diodes. The soft preavalanche breakdown is mainly due to the effects of electron-hole generation and recombination, intrinsic defect generation such as point defects during laser doping and surface leakage currents. From the values of the observed leakage currents, it is clear that the defect generation due to laser doping is minimal. Tian et. al. (2006) have also shown minimal defect generation due to laser doping by carrying out TEM and Rutherford backscattering analysis of laser-doped
Schottky diodes based on 6H-SiC substrates. However, an abrupt reverse breakdown voltage of about 12 V was observed for the as-received substrate (Figure 3.12a).
Appendix C shows the behaviour of various contact electrodes to 6H-SiC (n-type substrate).
98 CHAPTER 4: MODELLING OF DIFFUSION COEFFICIENT
4.1 Laser enhancement of diffusion
P-type doping of SiC and many other wide bandgap semiconductors still requires further understanding. Al and B are the most commonly used p-type dopants in SiC and epitaxial doping or ion implantation is widely used for incorporating these dopants into
SiC. Boron out-diffusion and complete activation are problems with both of these dopants incorporated by ion-implantation or epitaxial doping [Panknin (2001)]. Cr has been shown to serve as a p-type dopant in SiC [Achtziger (1998), Lebedev (1999)].
However low solubility limit (3x1017 cm-3), unavailability of dopant precursor in suitable chemical form and intricacy of doping with conventional techniques restricted the use of
Cr as an active dopant in SiC [Harris (1995)].
Laser doping provides an alternative technique to dope SiC with Cr. Salama et al.
(2003) and Tian et al. (2006) investigated this technique to dope SiC with n-type (N) and p-type (Al) dopants. Tian et al. (2006) showed that there are two distinct diffusion regions, near-surface and far-surface regions, which were identified in the dopant concentration profiles, indicating different diffusion mechanisms in these two regions.
The effective diffusion coefficients of nitrogen and aluminum were determined for both regions and found to be 2.4×10-5 and 9.2×10-6 cm2/s in the near- and far-surface regions for nitrogen, respectively, and 1.2 ×10-5 and 1.3×10-6 cm2/s in the near- and far-surface regions for aluminum, respectively. Their calculated diffusivities were at least six orders of magnitude higher than the typical values for nitrogen and aluminum, which indicated that the laser doping process significantly enhances the diffusion of Al and N dopants in silicon carbide. They assumed the temperature to be uniform within the depth of thermal
99 diffusion length and thus the diffusion coefficient was assumed to be independent of the depth.
During the laser doping process, however, the wafer is under a non-isothermal condition with rapid localized heating and fast cooling. The wafer temperature can be raised to sufficiently high temperatures in a very short time without exceeding the peritectic temperature, 3100 K, of SiC. The rapid heating causes unidirectional diffusion of the dopant atoms downwards in the substrate when the laser pulse is on. When the
Laser Beam
r0
Dopant Gas Ambient Concentration gradient
z = 0
Thermal Stress
SiC wafer E. M. F.
Focal plane (Beam waist) z
Figure 4.1. The mechanism of dopant diffusion due to the concentration gradient, laser electromagnetic field (E. M. F.) and thermal stresses in SiC during laser doping.
100 laser pulse is off, however, the atoms remain locked in their position for all practical purposes due to rapid cooling. These locked atoms diffuse further downwards when the laser pulse is on again. (Figure 4.1)
This research extends the work of Tian et al. (2006) by considering the diffusion of dopants in non-uniform temperature fields. The Fickian diffusion, thermal stresses between the laser-heated region and surrounding cooler region, and electromagnetic field of the laser beam contribute to the increased overall mass flux and solid solubility for even such a large atom as Cr in SiC (Figure 4.1).
4.2 Theoretical model
The diffusivity, D (T (z, t)), of Cr in SiC is assumed to be a function of temperature given by
⎛ − Q ⎞ ⎜ ⎟ = DtzTD 0 exp)),(( ⎜ ⎟ (4.1) ⎝ B tzTk ),( ⎠ where D0 is the pre-exponential diffusion constant, Q is the activation energy required for diffusion and kB is the Boltzmann constant. The total mass flux (J) of Cr in SiC is considered to be due to three effects, concentration gradient, thermal stresses and electromagnetic field of the laser beam, i.e.,
sF ++= JJJJ l , (4.2) where JF, Js and Jl are the Fickian mass flux and mass fluxes due to the thermal stress and laser field respectively. The Fickian mass flux is related to the concentration gradient by
tzdC ),( the expression −= DJ for one-dimensional diffusion where C (z, t), which is F dt measured in units of the number of dopant atoms per unit volume in this study, is the
101 dopant concentration at any depth z from the substrate surface at any time t. The mass flux Ji due to the driving force Fi, i = s, l in this case, is given by [Shewmon (1963)]
⎛ tzDC ),( ⎞ ⎜ ⎟ J i = ⎜ ⎟Fi . (4.3) ⎝ b tzTk ),( ⎠
The driving force due to the thermal stress can be expressed as
σAF == εE AsYss = + αe ΔT )(1 EA Ys , (4.4) where σ is the thermal stress, As is the laser-irradiated area at the substrate surface, ε is the thermal strain, EY is the Young modulus of elasticity, αe is the thermal expansion coefficient and ΔT is the temperature difference given by T(z)-T0. The driving force due
to the electromagnetic field of the laser beam is given by = eEF ll , where e is the electronic charge and El is the electromagnetic field of the laser beam given by [see
Appendix D],
2IU El = (4.5) cε0 where c is the speed of light and ε0 permittivity in vacuum.
Now the detailed expression for the total mass flux can be written as
⎡ ∂ tzC tzTD )),((),( tzTD )),(( ⎤ ⎢−= tzTDJ )),(( + ()eEtzC l α e Δ++ s (),)),(1(, EtzCAtzT Y ⎥ ⎣ ∂z B tzTk ),( B tzTk ),( ⎦
(4.6)
This expression for the total mass flux is used in the following mass conservation equation:
∂ tzC ),( ∂J −= (4.7) ∂t ∂z
Eq. (4.7) can be expressed as follows using the finite difference approximation, i.e.,
102 ()− ()0,zCt,zC ()(− t,zJt,zJ ) −= b (4.8) t − zz b
Here zb is the depth where the Cr concentration is equal to the background concentration of Cr in the parent wafer. ()zC 0, = 0 since there was no Cr in the parent wafer. The dopant concentration does not change after the depth zb and, therefore, the mass flux can be taken as zero at the depth zb, i.e., J(zb,t) = 0. Applying these two conditions to Eq.
(4.8) and noting that the dopant concentration profile is obtained from the doping experiment under the quasi-steady state condition, Eq. (4.8) is rewritten as follows
()t,zC ()t,zJ b −= b (4.9) t b − zz b where tb is the laser-substrate interaction time when the doping process is considered to attain the quasi-steady state condition. The interaction time tb is given by tb =2r0/v, where v is the speed of the laser beam relative to the substrate.
Combining Eqs. (4.2) and (4.6), the temperature dependent diffusivity can be expressed as
b − zz (),tzC b tb tzTD b )),(( = (4.10) ⎡ ∂ tzC b (),),( eEtzC lb α e Δ+ b (),)),(1( EAtzCtzT Ysb ⎤ ⎢− + + ⎥ ⎣ ∂z tzTk bB ),( tzTk bB ),( ⎦
Eq. (4.10) is used to determine the diffusivity and then Eq. (4.1) is applied to calculate the pre-exponential diffusivity and the activation energy for diffusion.
4.3 Experimental results
Since the thermal properties of the substrate dominate the conduction mechanism,
103 the temperature drops instantly to ~2300 K and ~2790 K for 6H-SiC and 4H-SiC respectively as shown in figures 2.3? and 2.4?. This non-equilibrium thermal condition induces non-equilibrium mass flux leading to an enhanced diffusion coefficient during laser doping. The dopant concentration profiles obtained by the SIMS analysis (Figure.
3.2a and 3.2b for 6H- and 4H-SiC respectively) and the temperature distributions
(Figures 2.3? and 2.4?) calculated using the thermal model are utilized to determine the concentration and the concentration gradient for different temperatures at different depths. To avoid the SIMS instrumental fluctuations in the concentration, the SIMS data were fitted to the following ninth order polynomials:
) ) 19 3 3 10 19 10 6H-SiC (n-type) 4H-SiC (p-type)-Cr 18 Polynomial Fit 10 Polynomial Fit 1018 1017 1017 1016
1016 1015
15 1014 (a) 10 (b) Cr Concentration (atom/cm Concentration Cr Cr Concentration (atoms/cm 0 200 400 600 800 1000 1200 1400 1600 0255075 Depth (nm) Depth (nm)
Figure 4.2 Curve fitted SIMS concentration profile for Cr along the depths of (a) n-type
6H-SiC and (b) p-type 4H-SiC wafers.
19 17 215 312 49 57 ()tz,C b z z z z ×−×+×−×+×−×= 1016.11044.81090.31015.11009.21095.1 z 64 zz 7 − z 83 ×−×+−×+ 102106.141.51002.1 − z 97
for 6H-SiC, (4.11) and
104 19 18 217 316 415 513 ()tzC b z z z z ×−×+×−×+×−×= 1083.31025.11052.21025.31074.21039.1, z z 611 z 79 z 87 ×−×+×−×+ 1036.11016.51025.81026.7 z 95
for 4H-SiC (4.12) where the unit of z is nanometer (nm). These two expressions were used in Eq. (13) to calculate the diffusion coefficient D (T (z, tb)). Figures 4.3a and 4.4a show the diffusion coefficients of Cr in 6H-SiC (n-type) and 4H-SiC (p-type) wafers respectively.
Depth (nm) 1200 1000 800 600 400 200 0 -20 -7 2
/s) -10 D =3.27x10 cm /s 2 5x10 6H-SiC (n-type) 0 Q=1.61 eV -22 4x10-10
-10 -24 3x10 Y = A + B * X Parameter Value ln (D) 2x10-10 ------26 A -14.93252 B -18786.64757 1x10-10 (a) ------(b) -28 Diffusion Coefficient, D (cm Coefficient, Diffusion 2200 2400 2600 2800 3.3x10-4 3.6x10-4 3.9x10-4 4.2x10-4 4.5x10-4 Temperature, T (K) 1/T (K-1)
Figure 4.3. (a) Variation in the diffusivity (D) of Cr at different temperatures (T) corresponding to different depths of the n-type 6H-SiC wafer. (b) Linear fit of the ln (D) versus 1/T plot to obtain the activation energy Q and pre-exponential diffusivity D0 for the diffusion of Cr.
The diffusion coefficient is 4.61×10-10 cm2/s at the surface and decreases to 7.14×10-14 cm2/s at 1.3 μm for 6H-SiC. The diffusion coefficient for the 4H-SiC wafer is found to be 6.75×10-12 cm2/s at the surface which decreases to 6.83×10-13 cm2/s at a depth of 71 nm. The lower diffusivity of Cr in 4H-SiC may be attributed to the presence of
105 aluminum (1019 cm-3) in the wafer lowering the available sites for Cr to occupy.
The pre-exponential diffusion constant and the activation energy were determined by plotting ln(D(T(z, tb))) as a function of 1/T as shown in Figures 4.3b and 4.4b, and fitting a straight line through the data points. Usually the least square technique is used to fit a straight line through the data points. While the straight line passes through most of the data points for 6H-SiC in Figure 4.3b, very few data points have been chosen for the straight line for 4H-SiC in Figure 4.4b on physical grounds as explained below.
Depth (nm) 60 50 40 30 20 10 0 -11 /s) -25.5 2 1x10 Q = 4.873 eV 4H-SiC (p-type) -4 2 8x10-12 D = 5.84x10 cm /s -26.0 0
6x10-12 -26.5 Y = A + B * X 4x10-12
ln (D) Parameter Value ------27.0 -12 A -7.44597 2x10 (a) B -56551 (b) ------Diffusion Coefficient, D (cm -27.5 3010 3020 3030 3040 3.28x10-4 3.29x10-4 3.30x10-4 3.31x10-4 Temperature, T (K) 1/T (K-1)
Figure 4.4 (a) Variation in the diffusivity (D) of Cr at different temperatures (T) corresponding to different depths of the p-type 4H-SiC wafer. (b) Linear fit of the ln(D) versus 1/T plot to obtain the activation energy Q and pre-exponential diffusivity D0 for the diffusion of Cr.
The diffusivity is expected to be constant in the small doped region (~75 nm in Figure
4.2b) over which the temperature variation is small (~28 K in Figure 4.2b). A small set of calculated values of ln(D) is found to be constant over a certain range of temperature
106 in Fig. 4.3b, which has been selected for linear fit. This line also represents an average value of all the values of ln(D) because some of the values are above and below this line.
Comparing the equation of the fitted straight line to the following linear expression of the diffusion coefficient
− Q ⎛ 1 ⎞ tzTD b ))),((ln( = ⎜ ⎟ + D0 )ln( , (4.13) B ⎝ Tk ⎠ the activation energy (Q) and the pre-exponential diffusion constant (D0) were calculated.
-7 2 Q = 1.61 eV and D0 = 3.2×10 cm /s for the 6H-SiC wafer, whereas Q = 4.873 eV and
-4 2 D0 = 5.83×10 cm /s for the 4H-SiC wafer corresponding to lower diffusivity of Cr in
4H-SiC.
Since very limited data are available in the literature for the diffusivity of Cr in
6H- and 4H-SiC, the diffusion coefficient was compared to the volume diffusivity of 3C-
SiC. Takano et al. (2001) studied the diffusivity of Fe, Cr and Co in chemical vapor-
-14 2 deposited 3C-SiC and reported the diffusivity as 2.3×10 cm /s at 1573 K with D0 =
9.5×10-11 cm2/sec and Q = 81 KJ/mol (i.e., Q = 0.84 eV) for Cr. Based on the estimated
-12 values of D0 and Q in present work, the diffusion coefficient of Cr in 6H-SiC is 2.2×10 cm2/s at 1573 K, which is almost two orders of magnitude higher than the reported value
[Takano et. al. (2000)], indicating an enhancement of diffusivity in laser doping.
The mass fluxes of Cr in the wafer are presented in Figure 4.5 to examine the effect of individual driving forces on the diffusion process. In the case of 6H-SiC, it is observed that the mass flux due to the laser electromagnetic field is larger than that due to the concentration gradient. For the 4H-SiC wafer, however, the mass flux due to the concentration gradient is larger than that due to the laser electromagnetic field. This is
107 because the absorptivities of 6H-SiC and 4H-SiC are 13.5% and 70.3% respectively.
These are actual measured values and the large difference in the absorptivity is due to the difference in the types of sample. The n-type 6H- SiC wafer was very transparent and the p-type (Al-doped) 4H-SiC wafer was very opaque. Therefore a higher irradiance
(~2.17×108 W/cm2) was used for doping 6H-SiC than the
16 16
10 10 E.M.F Concentration Concentration ·s) ·s) E.M.F 2 13 Stress 2 13 10 10 Stress
1010 1010
7 107 10
4 4 10 (a) 10 (b) Mass Flux (atom/cm
Mass Flux (atom/cm 1 101 10 80 0 200 400 600 800 1000 1200 0 20 40 60 Depth (nm) Depth (nm)
Figure 4.5 Diffusive mass fluxes of Cr due to three types of driving forces: concentration gradient, laser electromagnetic field (E. M. F.) and thermal stresses for (a) n-type 6H-SiC wafer and (b) p-type 4H-SiC wafer.
irradiance (~3.2×107 W/cm2) used for 4H-SiC. The type and concentration of dopants in the parent wafer and the effect of temperature on the absorption coefficient can further attenuate the beam, reducing the irradiance of the laser beam as it propagates through the wafer and thus generates different electromagnetic fields in different wafers.
Additionally, the solid solubility was higher for 6H-SiC than for 4H-SiC. The mass flux due to thermal stresses is low in both cases due to the low thermal expansion coefficient of SiC.
108 Table 4.1 summarizes the results of diffusion coefficient obtained for various dopants in
SiC during laser doping. It provided a very good comparison to the conventional doping techniques. In case of Cr, no such data is available for conventional doping and therefore this information provided a good baseline for Cr doping.
Table 4.1.Comparison of diffusion coefficient for various dopants in SiC under conventional doping and laser doping.
Wafer Dopant Conventional Doping Laser Doping Diffusion Diffusion Temperature Temperature coefficient coefficient (°C) (°C) (cm2/s) (cm2/s) Al 3×10-14-1.2×10-12 1800-2300 3×10-7 ~2800 SiC N 2×10-13-5×10-12 1800-2450 3.9×10-7 ~2800 Cr NA NA 3.27×10-7 ~2900
The model presented in this chapter can be utilized to estimate the diffusion coefficient for any dopant element doped using laser doping technique. It provides an advantage for extraction of the information from the experimental data which incorporates all the effects that are difficult to consider during theoretical model work.
109 CHAPTER 5: SIC WHITE LIGHT EMITTING DIODES
5.1 Indirect bandgap SiC
SiC is an indirect bandgap semiconductor, therefore, SiC LEDs are generally less efficient than direct band gap GaN and GaAs based LEDs. However, efficient luminescence has also been discovered in silicon, which is an indirect bandgap semiconductor, due to its defect states. The strain in the Si lattice shifts the conduction band to form a direct band gap material [Wai et. al. (2001)]. It has been shown in GaP, an indirect band gap material, that by introduction of iso-electronic defects like nitrogen or arsenic it can be converted to a direct band gap semiconductor [Casey et. al. (1978) and
Lee et. al. (1985)]. Similarly, the impurity states (donor, acceptor) formed due to various dopants in SiC may allow electronic transitions similar to that of a direct bandgap semiconductor. Recent patent by Pan J. (2006) further simplifies the use of indirect band gap semiconductors using deep levels.
The low efficiency is partly compensated for by the ability to drive the SiC LEDs at higher currents. Kamiyama et al. (2006) considered the donor-acceptor pair (DAP) recombination mechanism for light emission from an N- and B-doped SiC LED and reported its internal quantum efficiency at 250 K to be 95% of what was obtained for
GaN at 10 K, indicating that DAP recombination is a promising mechanism for light emission in SiC. This DAP emission mechanism may have promise for fabricating efficient white LEDs. Additionally, Vlaskina (2002) demonstrated a SiC green LED having the same brightness as the Cree Research, Incorporated GaN-based LEDs, but higher stability and reliability over a wide temperature range.
110 5.2 Mechanism for White Light Emission in SiC
SiC is a compound semiconductor formed by combination of Si and C which belong to group IV. Therefore, Group III and lower elements act as acceptors while
Group V and higher elements act donors. These donor and acceptors can form one or more impurity levels within the forbidden bandgap of SiC and can contribute a single or multiple electrons or holes depending upon the group they belong. The emission of different colors to ultimately generate white light in SiC is tailored on the basis of these
DAP recombination mechanisms for luminescence.
Figure 5.1. Energy levels of multiple dopants in 6H-SiC for DAP recombination mechanism for RGB emission.
Figure 5.1 shows a DAP recombination mechanism for 6H-SiC n-type with Cr and Al.
Lebedev (1999) has listed the impurity state energy levels for various types of dopants impurity atoms and its behavior as an acceptor or donor based on the lattice site it occupies. However, slight variations in these reported impurity levels have been
111 observed by others depending on the dopant concentration, doping technique, measurement technique and substrate characteristics.
5.2.1 Single donors/acceptors (N / Al, B)
Nitrogen, a group V element is the main impurity atom in the as-received single crystal 6H-SiC wafer substrate. Nitrogen acts as a single donor by contributing one electron and produce the following shallow donor levels in the forbidden bands of 6H-
SiC: 0.17 eV (D1), 0.20 eV (D2) and 0.23 eV (D3) corresponding to three nonequivalent positions of carbon atoms in the lattice. Similarly it forms two donor levels in 4H-SiC:
0.052 eV (D1), and 0.092 eV (D2). [Krasnov et. al. (1969)]. The nitrogen atoms can also form excitons [Kholuyanov (1969)] which may be captured by neutral or singly ionized nitrogen, forming three-particle or four-particle nitrogen exciton complexes. Direct transitions of electrons from the shallow levels to the valence band or from nitrogen exciton complexes may also occur.
Aluminum belongs to group IIIA of the periodic table and thus acts as a compensating impurity, for 6H-SiC (n-type) wafer. However, it is the main existing dopant for 4H-SiC (p-type) wafers. It acts as a single acceptor and gives out a single hole upon replacing Si in SiC. Krasanov et. al. (1969) investigated the low temperature photoluminescence bands of 6H-SiC crystals containing different amounts of aluminum and reported three deep acceptor levels separated by 0.23 eV (A4), 0.39 eV (A3) and 0.49 eV (A2) from the valence band edge. Though Lebedev (1999) has reported these levels to be 0.23 eV (A3) and 0.1-0.27 eV (A2). In case of 4H-SiC it forms only one acceptor level 0.23-27 eV (A2). Their data indicate that luminescence may be possible in the range
112 of 2.3 to 2.8 eV due to the transitions of electrons between the impurity states, i.e, from the shallow levels to the deep levels. The observed blue luminescence component in these white LEDs in this study is mainly due to DAP recombination of electrons from N level with holes at Al-levels. [Bet et. al. (2007)]
Boron belongs to group IIIA and acts as single acceptor in SiC thereby contributing one hole. It forms deep energy levels of Ev +0.35 eV in 6H-SiC (A1) and Ev
+0.29 eV (A1) in 4H-SiC. Due to it smaller size it has a very good diffusivity, however easy out diffusion is also an equal problem during doping. Electron from N-level recombines with holes at B level to give blue green component in the White LED spectrum. [Bet. et. al. (2006)]
5.2.2 Double donors/acceptors (Se / Cr)
Chromium belong to group VIB and as discussed earlier in the DLTS section, it acts as a more p-type dopant contributing two holes (double acceptor) along with deep acceptor levels of Ev +0.54 eV (A1) for 6H-SiC and Ev +0.74 eV (A1) for 4H-SiC and shallower energy levels of Ev +0.17 eV (A3) and Ev +0.15 eV (A4) in 4H-SiC. Electrons from N-level recombine with holes at Cr level to give green component in the White
LED spectrum. [Bet. et. al. (2007) and (2008)]
Similarly, selenium belongs to group VIA and has shown to behave as donor impurity in SiC. Recent work by Reshanov et. al. (2007) has shown that selenium acts as double donors in SiC. Similar to chromium, selenium will provide one additional electron per atom than conventional dopants producing less strain and defects in SiC. The energy levels formed due to selenium are shallow like nitrogen [Bet et. al. (2007)]; however no
113 values reported in the literature are available. Excess electrons due to double donors will enhance the radiative recombination probability when used in conjunction with double acceptors such as chromium. Current work provides preliminary results on Se doping in
SiC.
The transitions between these donor, acceptor and mid bandgap defect states formed due to various impurities were studied systematically using electroluminescence technique. The wavelength output and the corresponding energy of photon were correlated to impurity transitions relations available for the recombination between donor-acceptor and exciton states.
5.2.3 Flow of electron and hole for DAP recombination under applied bias
The electron and hole flow in case of SiC white LED’s is similar to the conventional p-n junctions. However, the tunneling of electrons and holes between the donor and acceptor states is more or less dictated by the applied bias. The junction physics defines the statistics on the recombination rate under certain applied bias conditions. Figure 5.2 shows the band diagram from 4H-SiC doped with Se, 4H-SiC with existing Al as p-type dopant and 4H-SiC with existing Al and laser doped Cr. The
Fermi energy level (Ef) in case of Se-SiC is close to the conduction band (CB), for Al-
SiC it is close to the valence band (VB) and in case of Al-Cr-SiC it is further close to the valence band due to its more p-type characteristics.
114 Se-SiC N-type Al-SiC P-type Al+Cr-SiC More P-type
Ec Ec Ec Ef
Ef Ef Ev Ev Ev
Figure 5.2. Position of Fermi-level for various regions of 4H-SiC (p-type-Al) white LED sample doped with Se and Cr.
Under equilibrium conditions, the Fermi level across the p and n regions of 4H-
SiC (p-type) white LED formed by doping Se and Cr will align as shown in Figure 5.3.
When a forward bias is applied across the junction, the electrons are injected from the n- side and holes are injected from the p-side.
Al-Cr-SiC P-type Al-SiC P-type
Se-SiC N-type
Figure 5.3. Alignment of the Fermi-levels under equilibrium conditions for p and n regions of 4H-SiC (p-type) white LED sample doped with Cr and Se
115
Small barrier Electrons out Large barrier
Electrons in +VE
ΔV ΔV
-VE Holes in Small barrier
Holes out Large barrier
Figure 5.4. Flow of electron and holes across the device under low or medium forward bias.
Therefore Fermi-level for n-type Se-SiC will move further upwards towards the conduction band and in case of p-type Al-SiC and Al-Cr-SiC it will move towards the valence band. This causes bending of the bands resulting in a potential barrier due to formation of the depletion region. Figure 5.4 shows the band-diagram for the device under forward bias conditions. The electrons have a large barrier across the p-n junction which causes them to tunnel from the donor states to the acceptor states or to the valence band. As one goes on increasing the bias the Fermi-level keep moving towards the respective bands thus causing more bending of the bands. The barrier for flow of electrons and holes between Al-SiC and Al-Cr-SiC almost vanishes at intermediate biases. At considerably high biases as shown in Figure 5.5, a stage is reached where in the electrons and holes can flow through the conduction band and valence band due to lack of any barrier. It also results in very less recombination. However, the barrier
116 Electrons in Easy glide Electrons out No barrier No barrier
+VE ΔV
-VE ΔV
Holes out No barrier No barrier Holes in
Figure 5.5. Flow of electron and holes across the device under very high forward bias.
potential across the junction is very high which leads to the emission of generally higher energy photon from the electrons recombining through shallow donor state to VB or from the conduction band to the acceptor states. Direct CB to VB recombination can also occur during this state which can result in band edge emission. Therefore, there is a threshold voltage that is required to cause recombination and to maintain a certain recombination rate. Beyond the stage shown in Figure 5.5 another mechanism described later in the section of high injection current becomes operative dictating the emission wavelength.
5.3 White LEDs in 6H-SiC (n-type-N) wafers with Cr and Al
To study the effect of Al in n-type 6H-SiC substrate, it was laser doped with Al and EL measurements were carried out shown in Figure 5.6. Experimental parameter for fabricating these devices is listed in Table 8. The blue electroluminescence observed is due to D2-A3 and D1-A2 transitions, i.e., the transitions from the N donor level (2.80 eV
(D2) and 2.87 eV (D1)) to the acceptor level (0.23 eV (A3) and 0.39 eV (A2)) [Bet et. al.
117
Figure 5.6. Electroluminescence spectrum, device structure and blue light emission observed from a blue LED fabricated on 6H-SiC (n-type-N) substrate by laser doping Al.
Figure 5.7. Electroluminescence spectrum, device structure and red-green light emission observed from a LED fabricated on 6H-SiC (n-type-N) substrate by laser doping Cr.
Under high injection the same green LED turns red due to a metastable mid bandgap defect states and quantum mechanical effect.
118
Figure 5.8. Electroluminescence spectrum, device structure and white light emission observed from a LED fabricated on 6H-SiC (n-type-N) substrate by laser doping Al and
Cr. The DAP recombination mechanism yields RGB emission which combine to form pure white light.
(2007)]. Similarly, to study the effect of Cr in n-type 6H-SiC, it was laser doped with Cr and EL measurements were carried out shown in Figure 5.7. The green luminescence corresponds to the transitions between N-Cr donor-acceptor states i.e. D1-A1 (2.29 eV).
The red luminescence was observed in both the cases and was mainly due to the metastable mid bandgap intrinsic defect states. Kholuyanov (1969) reported this behavior due to complex nitrogen excitonic states. Since R (Red), G (Green) and B (Blue) wavelengths were generated in 6H-SiC. A single chip white LED was fabricated by laser doping Cr and Al in 6H-SiC. A broad EL spectrum starting from 400 nm (band gap of
6H-SiC 3.06 eV) to 900 nm covering the entire visible spectrum was observed giving the
RGB combination to form a pure white light as shown in Figure 5.8.
119 5.4 White LEDs in 6H-SiC (n-type-N) wafers with B and Al
Boron being much smaller in size can lead to much higher concentration as compared to Cr. Therefore, Cr from 6H-SiC (n-type) white LED’s was replaced with boron. Initial study was performed to understand B behavior in 6H-SiC (n-type). 6H-SiC
(n-type) was laser doped with B and EL response was measured. Earlier work on boron doped 6H-SiC n-type has shown luminescence in the deep green to yellow wavelength range for this mechanism [Kholuyanov (1969)]. However for the observed wavelength as
Boron doped n-type 6H:SiC 1235 λ=507.27nm 1230
1225 λ=722.32nm itrary units)
1220
1215 B Intensity (arb Intensity 6H-SiC-N 1210
300 400 500 600 700 800 900 Wavelength (nm)
Figure 5.9. Electroluminescence spectrum, device structure and blue-green light emission observed from a LED fabricated on 6H-SiC (n-type-N) substrate by laser doping B.
shown in Figure 5.9 is 507.27 nm, the probable transition appears to be from the nitrogen donor level of 2.83eV to a deep acceptor level of 0.58eV. [Bet et. al. (2006)].
White LEDs were further fabricated using the same approach described above. Cr was replaced by boron. The entire sample was further oxidized in oxygen environment for 15 mins. to prevent any boron out diffusion at the surface during the temperature
120 2x10-10 6H-SiC (n-type) doped with B and Al
2x10-10
-10 W/nm 1x10
5x10-11
400 500 600 700 800 900 1000
Wavelength (nm)
Figure 5.10. Electroluminescence spectrum and device structure of a White LED fabricated on 4H-SiC (p-type-Al) substrate by laser doping B and Al.
rise during EL. Corresponding EL response is shown in Figure 5.10.
5.5 White LEDs in 4H-SiC (p-type-Al) wafers with Cr and N
Similar study was performed in 4H-SiC (p-type) wafers as well. To study the effect of N in p-type 4H-SiC substrate, it was laser doped with N and EL measurements were carried out. Experimental parameter for these experiments is also listed in Table 3.4.
A similar blue (460-498 nm) luminescence component in these white LEDs may be due to N-Al DAP i.e. D2-A3 and D1-A3 transitions, i.e., the transitions from the donor level
(3.148 eV (D1) and 3.108 eV (D1)) to the acceptor level (0.27 eV (A2)). [Bet et. al.
(2007)].
121
Figure 5.11. Electroluminescence spectrum, device structure and orange light emission observed from LED fabricated on 4H-SiC (p-type-Al) substrate by laser doping with Cr.
Figure 5.12. Electroluminescence spectrum and device structure of a green LED fabricated on 4H-SiC (n-type-N) substrate by laser doping Cr. More of red emission is observed due to a metastable mid bandgap defect states and quantum mechanical effect.
122
Figure 5.13. Electroluminescence spectrum, device structure and white light emission observed from a LED fabricated on 4H-SiC (p-type-Al) substrate by laser doping Cr and
N.
Similarly, to study the effect of Cr in p-type 4H-SiC, it was laser doped with Cr and EL measurements were carried out shown in Figure 5.11. Orange-red (677 nm) luminescence was observed. However, the mechanism this emission is different than the DAP mechanism and more detailed study is required to understand it further. Cr was also laser doped in 4H-SiC (n-type) substrate to study the Cr-N interaction. The green (521-575 nm) luminescence observed (Figure 5.12) corresponds to the transitions between N-Cr
DAP i.e. D1-A1 (2.29 eV). A corresponding red luminescence (698-738 nm) was observed in all the cases due to the same reasons as observed in case of 6H-SiC. A prominent violet wavelength (408 nm) was also observed under high voltage conditions and was mainly due to valence band to Al acceptor level transitions. This RGBV donor acceptor recombination as shown in Figure 5.13 showed broad EL spectrum of white
123 light starting from 380 nm (band gap of 4H-SiC 3.26 eV) to 900 nm covering the entire visible spectrum.
5.6 White LEDs in 4H-SiC (p-type-Al) wafers with Cr and Se
Since Se acts as a double donor and Cr being double acceptor, N from the 4H-SiC
(p-type) white LEDs was replaced with Se. Initially Se response was studied in 4H-SiC
(p-type) substrate by laser doping it. The corresponding EL response is shown in Figure
5.14. A violet-blue response was observed and this spectrum broadened with the increase in the current indicating that Se forms shallow donor levels similar to nitrogen. The light appeared more bluish white due to red emission observed along with the blue emission due to the same reasons explained earlier.
2.0x10-9 4H-SiC p-type doped with Se
1.5x10-9
-9 W/nm 1.0x10
5.0x10-10 4H-SiC-Al-Se Bluish white
400 500 600 700 800 900 Wavelength (nm)
Figure 5.14. Electroluminescence spectrum, device structure and bluish light emission observed from LED fabricated on 4H-SiC (p-type) substrate by laser doping it with Se.
124 6x10-10 4H-SiC p-type-Cr-Se 5x10-10
4x10-10
-10 W/nm 3x10
2x10-10 4H-SiC-Al-Cr-Se 1x10-10 White LED
400 500 600 700 800 900 Wavelength (nm)
Figure 5.15. Electroluminescence spectrum, device structure and white light emission observed from LED fabricated on 4H-SiC (p-type) substrate by laser doping it with Cr and Se.
This device was further doped with Cr to form another white LED structure. The double donor and double acceptor based LED device showed (Figure 5.15) a similar broad band white light emission in the EL response. Thus it shows that the predicted acceptor and donor levels are accurate and show the transitions equivalent to their energy levels. This technique thus provides a new way of donor acceptor transition mechanism for obtaining white light LED samples. Tuning of different color wavelengths is possible by adjusting the dopants type and concentration. Even the color temperature can be easily controlled by controlling the dopant concentration.
5.7 CRI and CCT for SiC White LEDs
SiC has shown the potential of improved CCT and CRI primarily due to its broad band luminescence. Luminescence mechanisms in SiC governed by DAP between
125 various donors and acceptors can be tailored to obtain the desired CRI and CCT. In present work, the color space tri-stimulus values according to 1931 Commission on
Illumination [CIE (1932)] are X = 0.3322, Y = 0.3320 and Z = 0.3358 for 6H-SiC n-type doped with Cr and Al. The CCT was 5338 K and CRI was 98.32. The dominant wavelength was observed between 555nm to 565 nm. The initial spectral power output of the LEDs was very low 1.5nW. Similarly, the color space tri-stimulus values were X =
0.3322, Y = 0.3320 and Z = 0.3358 for 4H-SiC p-type doped with Cr and N. The CCT was 5510 K, and CRI was 96.56. The dominant wavelength was 474.9 nm with the spectral power output of 14 nW. The CCT values are very close to the incandescent lamp
(or black body) and lies between bright midday sun (5200 K) and average daylight (5500
K) and the CRI values are extremely good close to the average daylight value of 100.
6H-SiC N-Al-Cr
Figure 5.16. Laser-fabricated SiC white LED showing the color space tristimulus values as per 1931 CIE chromaticity at 2 degree on 6H-SiC (n-type-N) wafer substrate laser doped with Al and Cr. Point A corresponds to the white light obtained from the emitted
RGB combination in comparison to the normalized reference point while point B represents the dominant wavelength (565 nm) in the emitted light output.
126 4H-SiC N-Al-Cr
Figure 5.17. Laser-fabricated SiC white LED showing the color space tristimulus values as per 1931 CIE chromaticity at 2 degree on 4H-SiC (p-type-Al) wafer substrate laser doped with N and Cr. Point B represents the dominant wavelength (474.9 nm) in the emitted light output.
Figure 5.16 and 5.17 shows the CCT and CRI values for 6H and 4H-SiC white LEDS respectively. CCT for the other two white LED’s was in the similar range and has been summarized along with the spectral power output. CRI for all these white LED’s was in the range of 85-99 %.
5.8 Power Output of SiC White LED’s
Power output for SiC LED’s is comparatively low and there are numerous factors which account for this. One of the primary goals of this work was to improve the power output at the same time maintain the color temperature to 5500 K for these white LED’s.
Different device designs along with several different contact configurations were utilized to achieve the goal. The preliminary results for the first three white LED’s have been
127 tabulated in Table 5.1. It is observed that the power output is in the nW range with an electrical to optical conversion efficiency of 10-4 to 10-6 %.
Table 5.1. Preliminary results on SiC White LED characteristics
Sample Dopants Device Contacts Dominant Peak Color Power substrate structure wavelength wavelength Temp output (nm) (nm) (K) (nW) 6H-SiC Al, Cr Al, W 565.7 697.3 5338 1.5 (n-type) probe (20 N-doped μm dia.)
4H-SiC Cr, N Al, W 554.6 474.9 5510 14 (p-type) probe (20 Al-Doped μm dia.)
4H-SiC Cr, Se 1.25 mm 589 615 6968 21.8 nW (p-type) Ag pin (483.2) (1100) (189 nW) Al-Doped and Al film
128 CHAPTER 6: FACTORS AFFECTING THE PERFORMANCE OF SiC WHITE LEDS
During scaling up of the power output of these LED’s at the same time maintaining the color temperature following factors are required to be taken into account:
1. Device structure and contact metal type (work function)
2. Contact configuration (area and size of electrode)
3. Color temperature tuning (dopant)
4. Power scaling
6.1 Device structure, contact metal type and contact configuration
Since all these parameters are interdependent on each other, an optimum blend of all is required to obtain higher power output. They also play important role in shifting the color temperature to the desired range. Different device designs with combination of different contact metal and contact configurations and their combined response of EL characteristics was studied and is tabulated in Tables 5.1, 6.1 and 6.2. The effect of contact metals on the I-V characteristics of the device has been studied and reported in
Appendix C. However these contact metal elements can also diffusion on the surface of the device during operation leading to change in the EL response of the device. Figure
6.1 shows the different contact configurations that were used during this study. Al foil and Al plate were utilized to primarily to increase the injection area and thereby reduce
129 Device Device Device Device Device Device Device
Ag Pins and 2 W probes 4 Ag Pins 2 Ag Pins Al foil and Al foil and Al foil 2 W Probes 8 Ag Pins Cu posts
Figure 6.1. Different contact configurations across the device with Al foil, Al plate,
Silver (Ag) pins, Copper (Cu) posts and Tungsten (W) probes.
the contact resistance. They also served as good thermal conductors to extract the heat generated during high injection. Cu post with 1 mm diameter upto 5 mm diameter were tested in this study and were again had the same purpose as Al. Due to oxidation and thermal degradation effects on these two contact electrodes, a more noble Ag pins with a
1 mm head and 20 μm shank was used as well. Tungsten probes with 20 μm tip were also used as electrode. Under high injection current, 1 mm diameter thick Ni wire was also used across these devices. As seen from the Table 6.1, voltage is comparatively higher than typically used, and is usually equal or slightly higher than the band gap of the semiconductor used for LED fabrication. In the present study, this voltage is comparatively higher and is mainly due to the parent wafers electrical properties. There was a shift observed in the color temperature with the change in the operating voltage and current for different contacts and contact configurations. Obtaining an optimum condition for a decent power output and at the same time very good color temperature to 5500 K was achieved in this work. This is indicated by the rows highlighted in yellow. However,
130 Table 6.1. Summary of SiC LED characteristics on 6H-SiC substrates with different contact metal and contact configurations.
Dominant Peak Color Power Sample Device I V Dopants Contacts wavelength wavelength Temp. output substrate structure (mA) (V) (nm) (nm) (K) (μW) Cr, Al, 574 1006.5 3545 1.29 160 32 Cu posts (590.3) (1017.6) (1890) (498) (275) (38) (1mm) Cr, AlAg 556.2 496 5405 2.91 600 22 pin and (571.2) (511.7) (4313) (25.5) (850) (23) Al foil 6H-SiC Cr, Al, 2 555.6 497.1 5432 12.4 780 27 (n-type) Al, Cr Ag pins (567.6) (498.9) (4600) (52) (900) (27) W-3 Cr, Al 4 N-doped Ag pins, 560.4 505.3 5174 3.05 950 19 Al foil
Cr, Al, 2 W probes, 562.4 507.1 5050 1.86 950 20 Al foil Cr, Al, Ag pin, 558.2 503.2 5277 9.2 800 26 Al foil Al, 2 Ag 505.6 1103. 6573 0.07 600 21 pins 6H-SiC 4 Ag pin, 501.4 491.4 7364 0.15 400 24 (n-type) Al Al foil 4 Ag pins, 2 512.5 497.7 6972 0.041 80 26 bias Cr, Al, 10 mm x 10 Ag pin 552 1105.8 5630 8.5 950 26 mm and foil
Au, Al, 521.2 498.3 6539 0.261 400 23 6H-SiC probe, (582.2) (1105.1) (2437) (59) (380) (35) (n-type) foil Al, Cr WL-3 (Flip) Oxide P on top and n 504.1 487.5 6636 0.142 200 33 probe, at bottom (595.7) (377.1) (0) (36.8) (500) (32) foil Cr, Al Ag 552.3 499.6 5590 11.07 850 24 pin, Al (558.6) (500.6) (5261) (19) (1000) (23) 6H-SiC B, Al, foil (n-type) oxide Cr-Au, N doped layer Al, Ag 545.9 502.9 5818 0.021 600 21 pin, Al (605.6) (1105.1) (0) (36.1) (800) (28) foil
6H-SiC Cr, Al, 514 384.3 7008 0.44 300 17 (n-type) Cr, Al Ag pin, (606) (382.6) (0) (2590) (600) (21) W-4 Al foil
131 Table 6.2. Summary of SiC LED characteristics on 4H-SiC substrates with different contact metal and contact configurations.
Dominant Peak Color Power Sample Device I V Dopants Contacts wavelength wavelength Temp. output substrate structure (mA) (V) (nm) (nm) (K) (μW) 4H-SiC Cr, Al, (n-type) Cr, N Ag pin, 585.7 696.8 2920 2.89 900 18 W-2 Al foil N doped 4H-SiC Cr, N Cr, Al, 472.2 422.1 20375 1.0 300 38 (p-type) Ag pin, (615.3) (461.9) (2739) (112.6) (500) (38) W-1 Al Al Foil doped Cr, Al, 580.1 917.7 3193 0.051 20 35 Cu posts (592.1) (1034.5) (0) (1914.5) (250) (30)
4H-SiC Al, Se Cr, Al, 2 599.5 571 4350 21.23 700 22 (p-type) probes, (616.1) (1105) (1903) (162.3) (999) (23) 4P-2 Al Foil Al doped 4H-SiC Cr, Se Cr, Al, 584.8 1105.2 2803 6.43 750 14 (p-type) Ag pin, (605.8) (1105.8) (0) (3125) (990) (11) 4P-1 Al foil
Al (Cr-Au, doped Al) for much higher output (~2.5 mW) one has to trade of the color temperature, which shifts towards 0 K or red under high injection. This is shown by the row highlighted in green.
Similar results have been presented in Table 6.2 for white LED’s fabricated on
4H-SiC substrates. The highest power obtained by sacrificing the color temperature was
~3.125 mW. However, this power is the total power from 380 nm to 1100 nm. If we consider only the visible range then the power is about 3-8 % of the total power and rest all is the near infrared (NIR) emission. The goal of obtaining high power in the visible still persists at the same time how to improve the color temperature. To solve this issue different methodologies were employed. Color temperature tuning and power scaling was handled separately.
132 6.2 Tuning of color temperature to obtain pure white light
From both the Tables 6.1 and 6.2, it was observed that under high injection the color temperature was shifting towards red. Therefore to maintain the color temperature it was necessary either to suppress the red/NIR emission or to increase the green or blue emission. Suppressing the red/NIR emission was slightly difficult as it was due to intrinsic metastable mid bandgap states and due to metallic behavior of SiC under high injection current. More detailed explanation for this emission is given in the later section on effect of high injection current on SiC LED performance. It leaves only one option to increase the blue and green emission. To how much extent one should increase this emission or rather what percentage of green or blue component is required to improve the color temperature from 0 K to 5500 K was the question required to be answered. For this two sets of experiments were performed
1. Use of green light source to determine the percentage of green component
required
2. Application of this information to achieve the desired color temperature in
devices
6.2.1 Color temperature tuning with a green laser
Figure 6.2 shows the experimental setup used for tuning the color temperature of the device using a green laser. A green laser source with 5 mW output, 7459 K color temperature and 532 nm output wavelength was used as a laser source.
133
Figure 6.2. Experimental setup for determining the % of green component required to improve the color temperature from 0 K to 5500 K
A 6H-SiC n-type wafer doped with Cr was the device used. This device was driven at high injection current of 900 mA at 9 V where the color temperature for the
Table 6.3. Measured values of the color temperature and power outputs for device, combined structure and green laser.
Green Laser Combined Structure 6H-SiC-N-Cr-2 Ag pins Color Color Color Output Output Output Temp Temp Temp V(V) I (mA) (μW) (μW) (μW) (K) (K) (K) 0.22 7459 316 3520 9 400 0.76 7472 450 4661 8 400 1.08 7473 642 4630 8 400 2.04 7481 168 6998 8 400 2.04 7481 139 6264 8 400 4.38 7407 783 5274 22 400 4.38 7407 126 7246 15 378 4.8 7467 235 7332 9 400 30.5 7463 306 7330 20 400 561 0 9 400 932 17 400
134 device was 0 K and power output of ~ 561 μW. The color temperature and the power outputs were measured using an integrating sphere. Green laser power was increased slowly while keeping the device power output constant. The device power output was changing during the experiment under same conditions of current and voltage, therefore total power was measured along with the green laser power output. This data was then plotted to see the shift in the color temperature of the device with the increase in the green laser power. Table 6.3 shows measured values of the experimental parameters and results. Figure 6.3 shows that only ~0.5-1 % of green component is required to improve the color temperature from 0 K to 5500 K.
8000 Green Laser (7463 K) 7000
6000 Standard reference white light (5500 K) 5000
4000
3000
2000
Color Temperature (K) Color Temperature 1000 Device (0 K) 0 0.00 0.02 0.04 0.06 0.08 0.10 Ratio (Green Laser output/Device output)
Figure 6.3. Graph of ratio of green laser power output to the device output vs. the color temperature. ~0.5-1 % of green component is required to improve the color temperature from 0 K to 5500 K
135 6.2.2 Color temperature tuning with two device approach
This information was utilized further to realize the improvement in the color temperature. Two different devices were used for the experiment. Each device was separately measured for its power output and color temperature and then a combined measurement was performed. This concept not only provided a solution for adjusting the color temperature but also provided an approach for scaling up of the power. Figure 6.4 shows the experimental setup and connection configuration for each of the devices.
Device A is a blue LED in 6H-SiC-N with Al and device B is a white LED in 6H-SiC-N-
Ga-Cr. Ga forms almost similar levels as Al and acts as a p-type dopant in SiC. Table 6.4
Figure 6.4 shows the experimental setup and connection configuration for color temperature tuning with two devices.
136 shows the values measured for these individual as well as combined devices. The information in Table 6.4 was plotted in two ways. Figure 6.5 shows the improvement in the color temperature by using a two device approach. While, Figure 6.6 shows that the scaling of the power non-linearly with the two device approach. The behavior for this non-linearity is explained in the later section on the effects of high injected currents in
SiC device performance. The basic conclusion that can be drawn from this study is that the color temperature can be easily tuned to the desired value by this approach. A further improvement can be to dope a tiny region surrounding the existing device to provide necessary color component to tune the color temperature.
Table 6.4. Summary of measurements for color temperature tuning with two devices.
Color Output Dominant Peak Device I (mA) V (V) Temp R (ohm) (μW) (nm) (nm) (K) A 200 19 0.094 8514 490.2 474 95 B 150-200 25 0.05 7825 497.8 1096.7 125 A + B 400 27-24 0.1308 7086 504.3 1105.8 60
A 400 25 0.248 8224 492.6 489 62.5 B 400 26-24 0.226 6275 527.1 1105.1 65 A + B 800 29-25 1.47 5902 543.2 1105.8 31.25
A 500 28-25 0.205 7419 497.7 485.4 50 B 500 24 0.548 5888 543.5 1105.8 48 A + B 1000 27-24 3.69 5311 558 1105.8 24
137 Current (mA) Device A+B 400 500 600 700 800 900 1000 9000 9000
8500 8500
8000 8000
7500 7500
7000 7000
6500 6500
6000 6000 Color Temperature (K) Color Temperature
5500 5500 Color Temperature (K)
200 300 400 500 Current (mA) Device A or B
Figure 6.5. Color temperature tuning with two device approach
Current (mA) Device (A+B) 400 500 600 700 800 900 1000
(24 V 5311 K) Device A Device B (25 V 5902 K) Device A+B 100 (24 V 5888 K) W) μ
(25 V 8224 K) (24 V 7086 K)
(24 V 6275 K) (25 V 7419 K) Power ( 10-1 (19 V 8514 K)
(25 V 7825 K)
200 250 300 350 400 450 500 Current (mA) Device A or B
Figure 6.6. Power scaling with two device approach
138 6.3 Power scaling of white LED in the visible regime
The second issue is the power scaling in the visible range. As discussed earlier power can be scaled by following ways:
1. High bias voltage with decent current (promoting more recombination for blue
and green emission)
2. High injection current
2.1. Provide greater interaction time for electrons to recombine with holes
2.2. Increasing the device area/volume
3. Two or multiple device approach
To utilize these parameters one need to understand the effects of voltage and injected current on the SiC LED output. Ultimately the key in achieving high power in visible range is in proper channeling of electron and hole for recombination in the impurity level created by various dopants to emit more white light.
6.3.1 Effect of high voltage on SiC LED output in the visible range
EL is only observed above a certain threshold voltage of ~ 17-20 V (voltage required to flow appreciable current) under forward bias conditions. No light emission is observed under reverse bias in case of these SiC LEDs suggesting that light emission is not due to the intrinsic type EL but due to injection type EL. [Futagi et. al. (1993),
Mimura et. al. (1994), Macfarlane et. al. (2001), Castagna et. al. (2003)]
In conventional LEDs, for tunneling injection to radiative recombination states, electrons are assumed to tunnel through the junction to conduction band tail on the p-side of the compensated diode. Usually a single impurity dopant is utilized to make p or n
139 10-6 4H-SiC-Al-Se
-7 10 500 mA 28 V 500 mA 36 V 500 mA 45 V 10-8 W/nm
10-9
10-10
400 500 600 700 800 900 1000 1100 Wavelength (nm)
Figure 6.7. Power scaling with the high voltage in the visible range under same high injection current
type. The density distribution of impurity tail states is approximated to be exponential.
Luminescence occurs only if the dynamic balance between the voltage dependent tunneling rate and the radiative recombination rate is satisfied such that the low energy tail states are saturated. However in case of SiC, DAP recombination mechanism is the prominent mechanism for light emission. In other words, emission of the electroluminescent radiation is due tunneling of electrons between impurity levels.
During this mechanism, electrons tunnel to these donor states or from the donor states to the acceptor states. Since multiple dopants are used to fabricate the p region, multiple acceptor levels are available. Applied bias dictates this tunneling phenomenon from one donor states to different acceptor states under low current injection. [Pan J. (2006)]
Therefore, it is observed that the output wavelength shifts with increase or decrease in the bias voltage. The wavelengths observed are smaller than the energy of the bandgap of the
140 host material. This is similar to the mechanism explained by Cohen (1975) for tunneling electroluminescent diode with voltage variable wavelength output. This allows tuning of the emission wavelength by controlling the applied bias. Other way to interpret this is one can choose a dopant material which can form specific impurity levels that can lead to desired emission of wavelength under certain bias conditions. As the injection current increases, more and more tunneling occurs which vanishes the bias effect due to filling of these states with electrons and the emission phenomenon becomes more random. Figure
6.7 shows the demonstration of this concept. Even under high injection, sufficiently high voltage can still cause transitions which can lead to emission in the visible region due to
DAP mechanism.
6.3.2 Effect of high injection current on SiC LED output in the visible range
The power output of SiC white LED in the visible range increases proportionally to the injected current only upto a certain range. Beyond this range, NIR emission starts dominating the spectrum. Figure 6.8 show the power scaling in 4H-SiC-Al-Cr-Se white
LED. At a constant voltage of 30 V, as the injected current is increased, a corresponding rise in power output is observed. The scaling is almost linear. No emission is observed in the NIR range indicating that all the transitions are purely due to DAP mechanism. A rise in temperature is observed upto ~ 200°C as shown in Appendix C. However, this temperature rise does not affect the wavelength output. The color temperature still remains in the range of pure white light.
If the injection current is now raised beyond 250 mA, one observes the dominating emission of red/NIR wavelength. The temperature of the SiC substrates measured at these currents (300 mA and above) reaches approximately 398-412°C. At
141 this stage an emission similar to the incandescent lamp is observed extending from 600 to
1100 nm with predominantly orange-red luminescence. The entire SiC wafer appears reddish orange in fraction of second as shown in Figures 5.7 and 5.11. If SiC was glowing red due to thermal effect then the temperature of SiC at this glow will be ~
1200°C or higher and at this conditions the contact Ag pins should have melted or vaporized. No such effect was observed on the contact pins. To further clarify this doubt,
SiC substrate was heated in a induction heater upto 620°C and was then further heated upto 1200°C in a batch furnace to see if it glows at that temperature. Figure 6.9 shows the results of these experiments. It was observed that SiC does neither glow at 620°C when heated in an induction heating setup nor at 1200°C in the batch furnace. A Ni-based
4H-SiC p-type-Cr-Se 1.25 mm Ag pin and Al foil 1E-9 80 mA 30 V (81 nW) 90 mA 30 V (96 nW) 100 mA 30 V (106 nW) 110 mA 30 V (115 nW) 150 mA 30 V (127 nW) 200 mA 30 V (189.9 nW) W/nm
1E-10
400 600 800 Wavelength (nm)
Figure 6.8. Power output scaling linearly with the injected current in the visible range upto 200 mA.
142 1200°C
Figure 6.9. Heating of SiC wafer substrate on induction heater upto 620°C and in a batch furnace upto 1200°C to determine if it glows at these temperatures.
superalloy which was introduced in the furnace at 900°C glows at 1200°C and therefore is almost invisible. A simple calculation for thermal emission for 4H-SiC (Eg =3.2 eV) at
-25 398 °C which is proportional to exp (-Eg/kT) comes to ~ 3.12×10 . This shows that there would be hardly any emission due to thermal effect.
To find out the exact operative mechanism at high injection both for temperature rise and NIR emission another experiment was performed. Here a 6H-SiC N-Al sample with oxide film on both sides and an Au-Ni film deposited on N side while an Al film deposited on p-side was subjected to cyclic injection currents. Figure 6.10 shows the sequence of measurements which is indicated by reading number. Table 6.5 provides the details of the experimental parameters and other characteristics of the light. A reference curve for incandescent lamp is also provided for comparison. At about 300 mA the device was reddish-orange as shown in the figure and visible along with NIR was observed in the spectrum. Keeping the device ON, the injected current was reduced to
143 100 mA where in the NIR emission disappeared and the power output in the visible range decreases correspondingly. Again the current was increased to 200 mA with device still in ON mode. The power output increased proportionately along with appearance of NIR emission. NIR emission was lower than that observed in case of 300 mA of current. Thus it showed that power even scales up proportionately in the NIR region along with visible.
The current was reduced again to 50 mA with device in ON conditions. The NIR emission disappeared again with power decreasing in the visible range proportionately.
This showed that the NIR emission was mainly due to metastable effects which are prominent after certain threshold of current. The color temperature change observed during these measurements shown in Table 6.5 is also a very good judge for these results.
Figure 6.10. Study of the observed NIR emission in SiC white LED under high injection current
144 The possible mechanisms for observance of NIR emission in SiC:
1. Quantum mechanical effect
2. Classical effect
6.3.2.1 Quantum mechanical effect
Quantum mechanically, during high injection the conduction and valence band broadens to accommodate the excessive injected electron and holes. This leads to the merging and to a certain extent overlapping of these bands. The electron hole recombination becomes similar to that observed in case of metal during incandescence.
As soon as the current is reduced, the conduction and valence band goes back to their original state due to emptying of the bands under high recombination rate which prevails from the previous high injection state. [Woodall et. al. (1998)]. This mechanism is similar to the Mott-Hubbard states causing reversible semiconductor-metallic transition in SiC.
[Ramachandran et. al. (1999)]. At low injection current, however, there are intrinsic metastable mid bandgap states formed within the bandgap of SiC and electrons can tunnel through these states leading to predominant orange-red luminescence. [Reshanov et. al. (2004)].
6.3.2.1 Classical effect
Classically, the high voltage which is present even during the high injection can cause impact ionization which can ionize more of the dopant atoms or the parent atoms. i.e. excite the electron from its ground state to the high energy states. These ionized atoms formed can further cause impact collisions again due to bias, leading to generation and excitation of more electrons. These excited electrons upon returning back to its
145 original ground states lead to the emission similar to that observed in the metals during incandescence.
Table 6.5. Experimental parameters and light output characteristics for SiC device during the excess injected current effect on SiC device output study.
Visible Color Total Measurement V I Dominant Peak Area Device power Temp power sequence (V) (mA) (nm) (nm) (mm2) (μW) (K) (μW) 6H-N- Reading 1 40-28 300 0.043 4389 713 570 6.5 Al, Reading 2 25-22 100 0.024 6779 503 489 0.038 Oxidize Reading 3 32-29 200 0.042 5383 556 1099 0.117 2.18 d, Au-N, Reading 4 22 50 0.01 6692 501 486 0.017 Al-P
6.3.3 Volume/Area effect for visible power output scaling
It has been already shown earlier that the power scales up with two device output which is indirect proof for scaling due to volume/area effect. An experiment was performed to prove this concept and then utilize it further for scaling using multiple devices. Figure
6.11 show the device and the connection configurations and Table 6.6 summarizes the experimental parameters and light output characteristics. The volume calculated is an approximate volume for the parallelepiped cylinder. Figure 6.12 clearly shows that the power output scales proportionally with the volume. The basic reason for this effect is larger volume corresponds to larger number of electrons and holes and thus greater recombination. Additionally, it give more time for the injected electrons to recombine with the holes as the hole mobility is less than that of electrons in SiC.
146 Table 6.6 Experimental parameters and light output characteristics for volume/area scaling
Visible Color Total Connection I power Temp Dominant Peak power Volume Area Configuration V (V) (mA) (μW) (K) (nm) (nm) (μW) (mm3) (mm2) Config 1 26 150 0.0277 6774 504.1 495.2 0.054 1.091 2.18 Config 2 26-23 150 0.0347 7500 497.6 490.6 0.06 1.296 5.9 Config 3 24-22 150 0.062 8202 492.2 480.7 0.111 4.749 9.54
Config 1 29-23 200 0.0262 6311 515.4 493.7 0.063 1.091 2.18 Config 2 28-23 200 0.0334 7293 498 486.9 0.064 1.296 5.9 Config 3 28-23 200 0.075 0 476 447.6 0.127 4.749 9.54
Figure 6.11. Device and connection configurations for the study of volume/area effect for scaling in SiC LEDs.
147 0.08
0.07 150 mA 200 mA 0.06
0.05 W) 380-780 nm μ 0.04
0.03
0.02 Power output ( 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Volume (mm3)
Figure 6.12. Scaling of the power output in the visible region proportionately with volume of the device.
6.3.4 Two or multiple device output for visible power output scaling
From the experiments on two device output scaling and volume/area effect for output scaling it is very clear that the power output scaling in the visible range is possible only if one is able to maintain enough voltage and at the same time tailor electron injection in effective manner to cause more recombination in the visible range. The connection configuration in two device output allowed lot of electrons to be lost to the Al plate at the bottom in the form of heat. Therefore, a two pin configuration with 1 mm diameter Ni –wire was used. Three different devices were used in this study. Each of these devices had six contacts of 1-2 mm2 area on each side. The p-side had Al film while
N-side had Ni-Au film. Figure 6.13 shows the devices used in this study and their connection configuration. EL responses of each of the devices was measured initially
148
Figure 6.13. Device and connection configuration for two or more device output for power scaling in the visible range.
80 6
V 70 Total (380-1100 nm) Power (mW) 4
6 5 Visible (380-800 nm) Power (mW)
V
60
3 6 4 50
40 3 V
V
0
8
7 30 4 2 20
V
V 1
V
10 0
8 3
5
2 0 2
Total (380-1100 nm) Power (mW) 0 Visible (380-800 nm) Power (mW) A B C A+B B+C A+C A+B+C Device at 300 mA
Figure 6.14. Total power output for single device, two devices together and three devices under constant current of 300 mA.
149 followed by two devices at a time. Finally response of all three devices together was taken. The current was held constant at 300 mA for all the measurements. Figure 6.14 shows the power output response for the devices alone, two together and all three together and their corresponding voltages. It is observed that the total power output of all three devices scales upto 63 mW in 380-1100 nm range and it is 3.66 mW in the visible range. With this two device or three device configuration one can clearly observe that the voltage is much higher almost twice or greater than the bias applied across single device.
This higher voltage promotes more recombination to occur in the visible range. Also now the total distance electron has to travel prior to being extracted out from the p-electrode is much greater. Since all the three devices are p-n junction diodes every single injected electron is going to witness three p regions prior to moving out of the system which gives much higher probability for it to recombine with the holes.
Therefore from all these power scaling experiments for visible output following conclusions can be drawn
1. Power scales up with the injected current.
2. High voltage bias is required for scaling visible emission under moderate
injection
3. Proper channeling of electron is necessary during high injection
4. Power scales up with two or multiple device approach
5. Electron-hole recombination can be promoted by tailoring a device design with
alternating p and n regions.
150 CHAPTER 7: SUMMARY 7.1 Conclusions
1. Laser doping allows effective incorporation of both conventional (Al, B, N) and
unconventional (Cr, Se) dopants.
2. Laser doping provides almost a defect free technique for simultaneous doping and
activation of dopants in SiC in a single step.
3. P-N junction diodes were fabricated for the first time on SiC using laser doping
technique.
4. Chromium (Cr) and Selenium (Se), which are unconventional dopants, were
successfully incorporated into SiC for the first time using laser doping.
5. The diffusion model presented in this work provides a convenient way of utilizing
the experimental dopant concentration profile to estimate the diffusion coefficient
and activation energy for dopant diffusion.
6. Dopant concentrations exceeding by a factor of 10-1000 than the solubility limit
were achieved in case of most of the dopants.
7. Two orders of magnitude increase in the diffusion coefficient was observed when
compared to the reported value at 1573 K in the literature.
8. Voilet, blue, blue-green, green, orange and red LEDs were fabricated in SiC using
different dopants and laser doping.
9. Color tuning approach was demonstrated.
10. White LEDs in both 6H-SiC and 4H-SiC have been successfully fabricated for the
first time in SiC using a laser doping technique.
11. SiC White LED provides good CCT and CRI
12. Demonstrated an approach for CCT tuning and power scaling.
151 7.2 Future Work
1. Tailoring of electron injection to avoid band broadening and overlapping effect
causing NIR emission. This can be achieved by either using a metal contact with
appropriate work function for by using a thin oxide layer on the injection (n-side)
of the device.
2. Converting the NIR to visible by use of up conversion phosphors. Even if the up
conversion efficiency for phosphors is very low (~10-20 %), it will still provide
significant boost in the visible range as 95% of photon emission in present white
SiC LED is in NIR.
3. Study the effect of high temperature caused due to high injection on contact
properties and device performance (Appendix C).
4. Dopant elements can be optimized further to add the effect of phonon generated
by them. Concept of phonon doping can be utilized to do this. (Appendix E).
Dopants such as Al, Cr, N can be replaced with Ga, Mg and Zn in incorporate this
effect at the same time get same EL response under DAP mechanism.
5. Optimized device design to allow tuning of color temperature. This can be
achieved by doping a ring surrounding the device with the dopant for desired
wavelength and providing options to activate the region when required for color
temperature tuning.
6. Optimization of electrical to optical conversion efficiency: Electrical conversion
efficiency in case of LEDs is defined as the ratio of number of photons emitted to
the number of electrons injected. In present research due to the thickness of the
152 substrate the voltage applied was quite high. Thinner substrates with optimum
device design can allow one to reduce the applied voltage and at the same time
achieve higher current injection without causing the band broadening effect.
7. Improve the extraction efficiency
a. Fresnel reflection and total internal reflection losses due to high refractive
index of SiC (n=2.55): Approximately 71% of the emitted light is lost due
to this effect. Device packaging using a graded refractive index epoxy can
serve to eliminate these losses to a greater extent. Additionally concepts of
perturbation of optical properties (Appendix F) can be used further extract
more light from the device.
b. Absorption losses due to thicker substrate or high dopant concentration:
Using thinner substrate with good device design will allow one to achieve
this.
8. Laser doping technique can be optimized and used for doping other
semiconductors and glasses (Appendix B and Appendix I)
9. Using the SiC white LED concept, solar cell can be fabricated in SiC and can be
tailored to absorb photons in the entire solar spectrum. (Appendix G)
10. Laser doping technique can be utilized to develop frequency selective surfaces
(FSS) for infrared detection applications. (Appendix J)
11. The concept of laser thin film deposition on flexible plastic substrate using Si
nanoparticles has been demonstrated and can be further utilized for small area
antenna fabrication. The concept can be applied to other materials systems and
applications as well. (Appendix H)
153
APPENDIX A: PROGRAM FOR X-Y SCANNING DURING LASER DOPING
154 Program
E, (Enter program)
C, (Control)
IA3M-0, (Initialize motor 3 to 0, direction along X)
IA1M-0, (Initialize motor 1 to 0, direction along Y)
S3M350, (Set speed for motor 3, ~ 600 divisons/sec correspond to 1 mm/sec)
S1M1000, (Set speed for motor 1, ~ 600 divisons/sec correspond to 1 mm/sec)
LM0, (locate the position for the start)
I3M-8000, (Specify the distance to move along X direction: motor 3)
P0, (Specify desired delay or pause)
I1M-1500, (Specify the distance to move along Y direction: motor 1)
P0, (Specify desired delay or pause)
IA3M0, (Specify the distance to move along X direction: starting location offset by Y)
P0, (Specify desired delay or pause)
I1M1460, (Specify the distance to move along Y direction: starting location offset by 40 divisions along Y)
P0, (Specify desired delay or pause)
L38, (Total number of loops depending on the area to be scanned)
LM-0, (Return back to the start position after the end of the loops)
R (Exit Program)
155
Area to be doped
Figure A.1. Velmex program for laser doping of SiC. Dotted region shows the area to be doped and each color shows the loop formed by the program and its progress for covering the entire area.
156 APPENDIX B : LASER DOPING OF GAP, SI, SIC WITH N, AL, PD AND B
157 Laser doping was also utilized to dope other elements such as Pd and B in
undoped SiC, Al in Si and N in GaP. Pd and B doping was carried out for fabrication of
SiC-based wireless optical sensors for temperature, pressure and ambient chemical
sensing. Figure B.1 shows the Pd counts spectrum in SiC and Table B.1 shows the
experimental parameters used for doping of the Pd. Figure B.2 shows the B counts
spectrum in SiC and Table e shows the experimental parameters used for doping of the B.
4x103
3x103
2x103 Pd
1x103 Counts per second
0 50 100 150 200 250 300 Depth (nm)
Figure B.1. Palladium dopant profile in laser doped 4H-SiC (undoped) substrate.
Table B.1. Experimental parameter for laser doping of Pd in undoped 4H-SiC
Sample Dopant Power Pulse Focal Spot # of Scanning Dopant medium (W) repetition Length size passes speed rate (mm) (μm) (mm/sec) (KHz)
4H-SiC Pd 8 CW 150 100 1 2 Tetrakis (Triphenyl (undoped) phosphine) Palladium Powder, argon 30 psi
Pd 12.5 5 150 80 1 0.8 Drive in, Argon 30 psi
158 1021
) 1020 -3
1019
1018
1017 Boron Conc. (cm
1016 0 20 40 60 80 100 120 140 160 Depth (nm)
Figure B.2. Boron dopant profile in laser doped 4H-SiC (undoped) substrate. The sample
has exceeded the solid solubility limit of 2.5x1020 cm-3 for B in SiC at the surface.
Table B.2. Experimental parameter for laser doping of Pd in undoped 4H-SiC
Sample Dopant Power Pulse Focal Spot # of Scanning Dopant medium (W) repetition Length size passes speed rate (mm) (μm) (mm/sec) (KHz)
4H-SiC B 11.9 5 150 100 1 0.7 Triethyl boron heated in (undoped) a bubbler to 80°C + argon (30 psi)
Laser doping of Si was carried to fabricate a tunable frequency selective surface. The
laser doping parameters and the SIMS concentration depth profile are listed in Table B.3
and Figure B.3 respectively. An additional study on the uniformity of dopant
159 concentration along the surface and at two different depths was performed. SIMS surface
scans were obtained at the surface, at a depth of 150 nm and at a depth of 300 nm.
Table B.3. Experimental parameters for laser doping of Al in p-type Si
Substrate Pulse Pulse Focal Beam Laser Scanning No. of Dopant Energy Repetition Length Spot Fluence speed Passes Source (mJ) Rate (mm) Diameter (J/cm2) (mm/sec)
(kHz) (μm) p-type Si TMA, 0.55 10 150 112 5.58 0.8 1 (100), ρ-1-10 Argon at
Ω-cm, 410 30 psi
μm Thick
Figure B.3. SIMS depth profile for Al concentration in laser doped p-type Si.
160 A Surface line scan 1E21 ~150nm depth line scan C ~300nm depth line scan
) D -3 1E20 B
1E19
150nm 1E18 Aluminum Concentration (cm Concentration Aluminum
0 200 400 600 800 1000 1200 1400 1600 1800 Scan Length (μm)
300nm
Figure B.4. SIMS surface scan along the beam path (AB) at surface, depth of 150 nm and depth of 300 nm.
Surface line scan ~150nm depth line scan ~300nm depth line scan 1E21 ) -3
1E20
A 1E19 C D 1E18 B Aluminum Concentration (cm 1E17 0 200 400 600 800 1000 1200 1400 1600 1800 Scan length (μm)
Figure B.5. SIMS surface scan across the beam path (CD) at surface, depth of 150 nm and depth of 300 nm.
161 From figures B.4 and B.5 it is observed that laser doping is quite uniform in both along
the beam path as well as across the beam path.
Laser doping of GaP with N was also carried out to fabricate red LEDs in GaP.
The experimental parameters and the EL response for the device is shown in Table B.4
and Figure B.6 respectively.
Table B.4. Experimental parameters for laser doping of Al in p-type Si
Sample Color Dopant Power Pulse Focal Spot # of Scanning Dopant Contribution (W) repetition Length size passes speed medium in LEDs rate (mm) (μm) (mm/sec) (KHz)
GaP Red N 6.7 15 150 100 1 0.8 Ultra high pure (undoped) nitrogen 30psi
1400 GaP doped with N 736.05621
Silver paint 1360 Laser doped N region 1320 GaP (undoped) 1280
Sliver paint 1240 Intensity (arbitrary units) 1200 400 500 600 700 800 900 1000 1100 Wavelength (nm)
Figure B.6. GaP deep red LED fabricated by laser doping N in undoped GaP and
corressponding EL response of the LED
This shows that laser doping technique is widely applicable to all materials and groups of
semiconductors.
162 APPENDIX C: STUDY OF CONTACT ELECTRODES TO SIC
163 C.1. I-V characteristics
Different metal contacts were used to study their effect on the I-V characteristic of the device. A 6H-SiC (n-type) with nitrogen concentration of 5×1018 cm-3 of thickness
444 μm was used in this study. The Schottky device structure is shown in Figure C.1.
Metal contact
n 6H:SiC (n-type)
Metal contact
Figure C.1. Schottky diode structure in 6H-SiC n-type wafer substrate.
The contact metal was deposited on a clean SiC substrate using thermal evaporation. Cu,
Au, Ni, and Indium were used as contact metals in this study. For one sample, the top surface was laser metallized using the same technique described by Tian et. al. (2006).
1.5-4 mm2 regions were deposited on the wafer surface. To study the effect of the contact area copper post with 1 mm diameter and sterling silver (95% Ag, 5% Cu) pins of 1 mm head and 20 μm diameter were also used along with the contact film. I-V measurement was performed with the probe station with 20 μm tungsten probes. The resistance measured for a bare SiC substrate without any contact metal with multimeter probes was
5904 Ω. Figure C.2 shows the I-V characteristics of the various contact metals and probes on SiC
164 Study of Contacts to SiC
30 8 6 20 4 10 2 0 0 -25 -15 -5 5 15 25 -2
Current (mA) -10 -4 -20 -6 -30 -8 Voltage (V)
Ni film with probes In Pressed with probes In soldered with probes Cu Film with probes Cu-Au with posts W Probes One side metallization Cu-Au Film with probes Cu Posts No Film
Figure C.2 I-V characteristics of the various contact metals and probes on 6H-SiC n-type wafer substrate.
Resistances calculated from the slope of the I-V curves are tabulated in Table C.1. It shows that laser metallized sample has the least resistance. This is because during laser metallization a carbon rich phase is formed on the surface which has excellent conductivity and thus the contact resistance is very minimal. Cu or Cu-Au film also have low resistance, however there are some issues with the adherence of Cu with SiC. Indium is a potential candidate since the work function of indium is very low giving good ohmic contact. Usually Ni is used as the contact metal for n-type SiC and Al is used for p-type as they form good ohmic contacts to SiC [Harris (1995)].
165
Table D1. Calculated resistance values from the I-V measurements for different contacts
Contact Type Resistance (Ω) One side metallization with W probes 2.8 Cu-Au film with W probes 138 Cu film with W probes 196 Indium pressed with W probes 207 Ni film with W Probes 228 Indium soldered with W probes 271 Cu-Au film with Cu post 975 W probes on bare SiC 2789 Cu posts on bare SiC 9813
C.2 Thermal effects during current injection
Temperature of the device increased with the increase in the injection current. A
6H-SiC n-type sample with Cu-Au film and 1 mm head Cu posts were used to perform this study. Figure C.3 shows the rise in the device temperature with time under constant current conditions. For currents upto 150 mA no change in the resistance is observed with time. Beyond 150 mA non-linearity is observed in the I-V characteristics which is mainly due to change in the surface properties due to high temperature effect. Figure 4 shows the calculated resistance values with time. The device reached maximum of 250°C at 250 mA in ~ 150 seconds.
166 6H-SiC n-type with Cu+Au Film 1 mm head
280 20 mA, 40 mA, 60 mA 80 mA, 100 mA, 120 mA 240 140 mA, 160 mA, 180 mA 250 mA C) ο 200
160
120
80 Temperature ( 40
0 0 100 200 300 400 500 600 700 Time (sec)
Figure C.3. Change in the device temperatures with time under constant current conditions.
1200 20 mA, 40 mA, 60 mA 1050 80 mA, 100 mA, 120 mA 140 mA, 160 mA, 180 mA 900 200 mA
(Ω) 750
600
450
300 Resistance 150
0 100 200 300 400 500 600 Time (sec.)
Figure C.4. Change in the total device resistance with time under constant current conditions.
167 C.3.Contact metal degradation during EL measurements under high injection
Degradation in the contact metal was observed under high injection current primarily due to thermal effects. This study was performed on a white LED sample on 4H-SiC p-type sample doped with Cr and N. 2 mm2 Cr contact metals were deposited as shown in Figure
C.5 on the p-region (Cr doped region) of the sample. Correspondingly Ni film was deposited on the N doped region to form N- contact. I-V characteristics of all the contacts were performed in the as deposited condition.
1 2 Cr film Cr film
4H-SiC (p-type) Cr-doped
3 4 Cr film Cr film
Figure C.5. Cr contacts deposited on p-region of the white LED sample.
Visual observation of the film under optical microscope showed a prominent degradation of the film. Extremely large grained structure with segregation along the grain boundaries
Cr contact as deposited Cr contact 1 after I-V 20X Cr contact 2 beside contact 1
Figure C.6 Degradation of Cr film under high current injection during EL measurements.
168 320 Contact 1 280 240 Before 200 After First Measurment After All Measurements 160 120
Current (mA) 80 40 0 0 3 6 9 12 15 18 21 24 27 Voltage (V)
Figure C.7. I-V characteristic of Contact 1
was observed. The effect does not remain on the contact where the current was injected but extends to the neighboring contacts as well. Figure C.6 shows the degradation of Cr
21 Contact 2 18
15 Before 12 After First Measurment After All Measurements 9
6 Current (mA) Current 3
0 0 3 6 9 12 15 18 21 24 Voltage (V)
Figure C.8. I-V characteristic of Contact 2
169 film under high currents for contact 1 where the current was injected and contact 2 which lies besides contact 1.
I-V characteristics of both the contacts were different. Figures C.7 and C.8 show the I-V characteristics for contact 1 and contact 3 respectively. After first measurement refers to I-V measurement after EL measurement at 50 mA of current and after all measurements refers to subsequent EL measurements upto 300 mA in steps of 50 mA.
From figure C.7 it is observed that the resistance has decreased as compared to as deposited condition to the condition after all measurements. However, Figure C.8 shows an exactly opposite trend. This may be due to formation of some oxides at the surface or some Cr compound at interface during the temperature rise due the high injection at contact 1. More chemical and materials analysis required to determine the cause and the effects observed on these films.
170 APPENDIX D:EXPRESSION FOR ELECTROMAGNETIC FORCE OF THE LASER BEAM
171 The irradiance, I(r, z), is related to the electromagnetic field, E(r, z), of a radiation by the following expression [Hecht (2006)]:
cε 2 z)I(r, = 0 zr,E )( , 2
where r and z are the radial and axial coordinates respectively and c and ε0 are the speed of light and permittivity in vacuum respectively. zr,E )( is the magnitude of zr,E )( , i.e., = * zr,Ezr,Ezr,E )()()( where E*(r,z) is the complex conjugate of E(r,z), and so
zr,E )( 2 = E2(r,z).
The reflectivity, transmissivity, absorptivity and absorption coefficient are
32.58%, 54.35%, 13.8% and 4.85 cm-1 for 6H-SiC and 26.6%, 3.37%, 70.3% and 67.71 cm-1 for 4H-SiC respectively. The skin depth, which is a reciprocal of absorption coefficient, is 2.06 mm and 147 μm for 6H- and 4H-SiC wafers respectively. The 6H-SiC wafer was 444 μm thick with light greenish transparent appearance, while the 4H-SiC wafer was 455 μm thick with deep bluish gray opaque appearance. Therefore the laser beam propagates through the entire thickness of the 6H-SiC wafer with very little absorption, while most of the laser energy is absorbed within the skin depth with very little transmitted energy in the case of 4H-SiC wafer. So the effect of the laser electromagnetic field would be confined over a smaller thickness of the wafer for 4H-SiC than for 6H-SiC.
Additionally the SIMS data indicate that the dopant depth is very low ranging from z = 80 nm for 4H-SiC to z = 1.3 μm for 6H-SiC. Over this small distance, the
172 change in the electromagnetic field with z would be negligible especially when a lens of large focal length (e.g., 150 mm in this study) is used to irradiate the wafer with the laser beam. So the electromagnetic force calculated at the wafer surface can be used for analyzing the effect of the laser field on the dopant diffusion inside the wafer. The variation of the electromagnetic force with the radial distance r is not considered because of the uniform irradiance profile of the laser beam and the one-dimensional thermal and dopant diffusion models. Therefore the electromagnetic force used in this study can be expressed as
2IU El = cε0
173 APPENDIX E: CONCEPT OF PHONON DOPING
174 The Debye temperature Θ is an important physical parameter of solids, which defines a division line between quantum-mechanical and classical behavior of phonons.
The Debye temperature depends on the dispersion law of acoustic phonons and, due to their complicated anisotropic spectrum, the analytic expression of Θ is not available in a general case. The Debye temperature can be estimated within the isotropic continuum approximation. Figure E.1 shows the Debye temperature for various elements of the periodic table while Figure E.2 shows it for compound semiconductors.
700
600
500
400
300
200
100 (K) Debye Temperature 0 Ge Si Ga As Al In Zn Se Zr Y Nb Pd
Elements
Figure E.1 Debye temperature for various elements of the periodic table
It is interesting to observe that most of the elemental or elements of the compounds forming direct bandgap semiconductors have the Debye temperature close to
175 1400
1200
1000
800
600
400
(K) Debye Temperature 200
ZnSe SiC GaNGaAs AlN InN GaP InP GaSbSiGe ZnO Compound Semiconductors
Figure E.2 Debye temperature for various compound semiconductors
the room temperature. And correspondingly indirect bandgap semiconductors elements such as Si or compounds SiC, AlN have exteremely high Debye temperature. This information provides an approach to improve the properties of indirect bandgap semiconductors for light emitting applications by doping them with elements which has
Debye temperature at room temperature. This element will provide more phonons which will assist in the indirect transition thereby improving the light emission efficiency.
To study this effect, SiC was doped with Ga instead of Al which has correspondingly higher Debye temperature. It was found that at same current and voltage conditions the output power for Ga doped SiC was higher than that of Al doped SiC.
However more detailed and systematic study is required in terms of the dopant concentration and dopant diffusion which will consequently effect the power output as well.
176 APPENDIX F: PERTURBATION OF OPTICAL PROPERTIES OF SIC
177 Tunable optical filters have been fabricated on silicon carbide wafer substrates by doping it with different dopants and applying bias. Undoped SiC, 6H: SiC (n-type) and
4H: SiC (p-type) wafer substrates were used for fabrication of these filters. Al, N, Sn, Ga,
Cr and Se were used as dopants.
The perturbation of optical properties due various dopants in SiC was studied by reflectivity and transmissivity measurements. Another goal of this work was to see if these large sized atoms cause the elimination for phonon tail which is observed in SiC due to its indirect bandgap characteristics. Figure F.1 shows the reflectivity and transmissivity measurements for various dopants laser doped in undoped SiC. Significant change was observed in these characteristics. Figure F.2 shows the disappearance of phonon tail for Cr and Se doped samples indicting that these large sized atoms can lead to
70 75 Sn Ga 70 60 Se Cr Undoped 65 50
60 Sn
Ga % Reflectivity
% Transmittivity 40 55 Se Cr Undoped 50 30 400 600 800 1000 1200 200 400 600 800 1000 1200 Wavelength (nm) Wavelength (nm)
Figure F.1 Transmissivity and reflectivity of various dopants in undoped SiC.
178 20 4H-SiC undoped, Doped with Cr, Doped with Se 6H-SiC (n-type), Doped with Cr and Al 20 16 Se -1/2 -1/2 15 Undoped Cr 12 N,Cr, Al cm cm N α) α) 10 8 Sqrt( Sqrt ( Sqrt
5 4
1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Energy (eV) Energy (eV)
Figure F.2. Indirect to direct conversion by using dopants such as Cr, Se and Al in undoped and 6H-SiC (n-type) substrate
disappearance of phonon tail which is nothing but conversion to direct band gap.
However, no significant change was observed in the absorption coefficient. In case of
6H-SiC white LED sample, the absorption coefficient increased in the visible range which is the good indication for white LED. However, the phonon tail was still present.
This lead to the evaluation of another aspect for SiC LED structure, i.e., is perturbation of the optical absorption coefficient and refractive index. The charge carrier concentration is controlled with the applied electric current at specific bias. It can thus modulate the reflectivity and transmissivity of the silicon carbide substrate.
Injection of current in the p-n structure alters the refractive index of the depletion region and thus affects the optical reflectivity of these various doped structures in SiC
(Figure F.3). This effect can be used as an optical filter. These properties can be tailored appropriately which right choice of dopants to improve the extraction or absorption of light for various energy conversion devices in SiC.
179
Figure F.3. Refractive index and absorption coefficient as a function of wavelength for n- type 6H-SiC substrate (5x1018cm-3) for different currents at fixed 15V bias.
180 APPENDIX G: SOLAR CELL APPLICATION-SIC
181 G.1 Photocurrent Measurement
Photocurrent measurements were carried out on two white light LED samples fabricated on C-1 (6H-SiC (n-type) laser doped with Cr and Al) and C-4 (4H-SiC (p-type laser doped Cr and N). 6H-SiC (n-type) and 4H-SiC (p-type) as-received substrates for white light LED samples were also characterized for comparison. A Keithley pico- ammeter/voltage source Model 6487 was used for carrying out the current measurements, while an Agilent digital multimeter Model 34401A was used for measuring the voltages and the resistances. The light source was a 50W tungsten filament lamp.
Light Source
Indium contact Laser doped Al p and Cr region n A 6H:SiC (n-type)
Indium contact
Figure G.1 Photocurrent measurement setup showing the light source, a pico-ammeter and Cr and Al laser doped 6H-SiC (n-type) white light LED sample.
182 Figure G.1 shows the photocurrent measurement setup. The p and n regions were fabricated by laser doping a n-type 6H-SiC (as–received nitrogen concentration of 5x1018 cm-3) and p-type 4H-SiC (Al doped) (as received aluminum concentration 1x1019 cm-3) wafer substrates with respective dopants. Cr and Al were used as p-type dopants while N was used as n-type dopant. A Q-switched Nd:YAG pulse laser (1064 nm wavelength) was used to carry out the doping experiments. For n-type doping, sample was placed in nitrogen atmosphere at pressure of 30 psi and laser-doped regions were formed on the sample surface by moving the chamber with a stepper motor-controlled translation stage.
The height of the chamber was controlled manually through an intermediate stage to obtain different laser spot sizes on the SiC substrate. . For p-type doping, TMA
(trimethylaluminum, (CH3)3Al) and Bis (ethyl benzene)-chromium were used. TMA was heated in a bubbler source immersed in a water bath maintained at 70°C until it evaporated and then a carrier gas, methane, was passed through the bubbler to transport the TMA vapor to the laser doping chamber. Similarly for chromium doping, Bis (ethyl benzene)-chromium was heated in a bubbler source immersed in a water bath maintained at 100°C until it evaporated and then a carrier gas, argon, was passed through the bubbler to transport the ethyl benzene chromium vapor to the laser doping chamber. These dopant gas species decompose at the laser-heated substrate surface producing Al and Cr atoms, which subsequently diffuse into the substrate. All samples were post-cleaned with a 45% by wt. KOH solution and then rinsed with acetone, methanol and D.I. water before making the electrical contacts. Aluminum and Nickel were deposited on p and n regions respectively on samples C-1 and C-4 prior to photocurrent measurements. Indium wire was also used as a contact for all four samples to see the change in the photocurrent due
183 to change in the contact metal. The values of the observed current and voltage in dark and light have been tabulated in Table G.1.
Table G.1. Photocurrent Measurements on C-1 (6H-SiC) and C-4 (4H-SiC) white light
LED samples and as received 4H-SiC (p-type) and 6H-SiC (n-type) silicon carbide wafer samples.
Sample Al-Ni contacts Indium Contacts
Dark Light Dark Light Resistance (KΩ) 4H-SiC (C-1) V (mV) 0.2 0.65 0.003 0.060 Al contact on p-side 8.619 and Ni contact on n- I (nA) 0.1 4.5 6 30 side 6H-SiC (C-4) V (mV) 0.002 0.013 0.006 0.028 Al contact on p-side 1.934 and Ni contact on n- I (nA) 0.28 3.8 30 55 side 4H-SiC (p-type) (as- V (mV) 0.45 39 2.4 88 7800 received) Al- contacts on both I (nA) 0.013 3.3 0.2 11.5 Overload after sides Al deposition 6H-SiC (n-type) (as- V (mV) 0.004 0.621 1.3 13.1 480 received) Ni contacts on both I (nA) 0.7 245 0.223 25 1.853 after Ni sides deposition
The mechanism for the observed photocurrent SiC is very simple, when a broadband source of light impinges on the doped SiC as received substrate or a SiC LED structure, the photons with different energies are absorbed, which creates electron hole pairs in the valence band. The photons with energies higher than the band gap of the substrate causes the electron to excite from the valence band to the conduction band where that electron is then flow out of the semiconductor through the contact wires to generate photo induced current and the corresponding voltage. A linear relationship
184 obeying ohms law is observed for the as received SiC wafer substrates while a non- linearity is observed in case of SiC LED samples due to presence of a p-n junction.
G.2 Concept for optimization of solar cell performance
Using the same DAP recombination approach used for white LEDs, the solar cells based on SiC can be designed and optimized to cover the entire solar spectrum. The high energy UV photon can also be easily tapped due to the broad band structure of SiC. The tunneling mechanism under high voltage bias can be utilized to prevent any loses that incur due the generated electron hole recombination. Additionally, the high bias will play an important role in accelerating the electrons towards the electrodes at a much faster rate.
185 APPENDIX H: LASER THIN FILM DEPOSITION ON PLASTIC SUBSTRATES USING SILICON NANOPARTICLES
186 Reduced melting temperature of nanoparticles is utilized to deposit thin polycrystalline silicon (c-Si) films on plastic substrates by using a laser beam without damaging the substrate. An aqueous dispersion of 5 nm silicon nanoparticles was used as precursor. A Nd:YAG (1064 nm wavelength) laser operating in continuous wave (CW) mode was used for thin film formation. Polycrystalline Si films were deposited on flexible as well as rigid plastic substrates in both air and argon ambients. The films were analyzed by optical microscopy for film formation, Scanning Electron Microscopy
(SEM) for microstructural features, Energy Dispersive Spectroscopy (EDS) for impurities, X-ray photoelectron spectroscopy for composition and bond information of the recrystallized film and Raman spectroscopy for estimating shift from amorphous to more crystalline phase. Raman spectroscopy showed a shift from amorphous to more crystalline phases with increasing both the laser power and irradiation time during laser recrystallization step. [Bet and Kar (2006)]
187 APPENDIX I: LASER DOPING OF GE124 QUARTZ SUBSTRATE
188 I.1 Physical and optical properties of GE124 quartz substrate
Table I.1. Physical and thermal properties of GE 124 Quartz substrates.
Sample Thickness Max. Use Thermal Thermal Absorptivity Density [μm] Temperature conductivity Diffusivity [%] [gm/cm3] [K] [W/cm.K] [cm2/s]
GE124 1210 1473 0.014 0.009 1.14 2.21
Reference: http://www.valleydesign.com/ge124-av.htm
100 4
80 % T 3 % R
% A % Absorptiion 60 @1064 nm % A = 1.1443 2 40
1 20 % Transmission & Reflection
0 0 800 900 1000 1100 1200 Wavelength (nm)
Figure I.1. Reflectivity and transmissivity measurements for GE 124 Quartz substrate.
I.2. Determination of temperature and laser doping parameters.
189
1600
1400 Pulse ON Pulse OFF 1200
1000
800
600
Temperature (K) 400
200 0 200 400 600 800 1000 1200 Depth (nm)
Figure I.2. Temperature distribution at the quartz surface and depth for laser beam spot size of 130 μm, average power of 10.5 W, pulse repetition rate of 5 kHz, pulse on time of
90 ns and velocity of 1 mm/s .
190 I.3. Laser Doping
Table I.2. Experimental parameters used during doping of n and p type dopants in GE
124 quartz substrate.
Sample Laser Dopant Power (W) Frequency Spot size Scanning Dopant medium Mode (KHz) speed GE 124 (μm) (mm/sec) Quartz
GS-1 Pulsed N 11.5 5 130 0.8 UHP Nitrogen (5 mm x 5 mm) Gas (30 psi) GS-2 Pulsed N 13.5 5 130 0.8
(5 mm x 5 mm)
GS-3 Pulsed N 15.5 5 130 0.8 (5 mm x 5 mm)
GS-4* Pulsed N 17.5 5 130 0.8 (10 mm x 10 mm)
GA-1 Pulsed Al 12.5 4.5 130 0.8 Trimethyl (5 mm x 5 mm) Aluminum organometallic GA-2 Pulsed Al 13.8 4 130 0.8 compound + UHP Argon (5 mm x 5 mm) (30 psi) GA-3 Pulsed Al 15.4 3 130 0.8 (5 mm x 5 mm)
GA-4* Pulsed Al 17 3 130 0.8 (10 mm x 10 mm)
UHP- ultra high pure, Aperture 3mm, 1064nm Nd:YAG laser, beam scan overlap of 60 % during rastering
over the region, * beam scan overlap of 40 %.
191 I.4. Hall Effect Measurements
Table I.3. Electronic and electrical properties of as received, nitrogen doped and aluminum doped GE 124 quartz substrate.
Sheet Thickness Concentration Resistivity Sample Current [nA] 3 concentration [μm] [/cm ] [Ωcm] 2 [/cm ] 1210 1.00 -2.870×109 8.873×108 -3.472×108 As received 0.05 1.00 -3.515×1013 3.615×104 -1.757×108 N doped 1210 1.00 -2.356×108 8.543×108 -2.850×107 Al doped 0.05 1.00 -5.044×1013 3.651×104 -2.522×108
I.5 Change in the absorptivity post laser doping
8 Al N 6 Undoped
5
3
% Absorptivity 2
0 800 900 1000 1100 1200
Wavelength (nm)
Figure I.3. Change in the % absorptivity for laser Al and N doped and undoped GE 124 quartz samples.
192 APPENDIX J: LASER DOPING OF SI WAFERS FOR FABRICATION OF TUNABLE FREQUENCY SELECTIVE SURFACES
193 J.1 Introduction
The spectral signature of a surface is defined by its radiometric properties. A signature is characterized by the way it reflects, transmits, absorbs or emits radiation over a wavelength region of interest. To shape the signature of a surface in the infrared, the surface or the media immediately surrounding the surface has to be physically altered.
Earlier work showed this feasibility using thin film approach and antenna array approach.
The former approach is a straightforward and well-known technique that incorporates optically-thin film coatings onto a surface. The latter technique requires complex design of a planar, periodic array of passive microantenna elements to cover the surface. Such arrays can function in both ways where the antennas are conducting elements or apertures in a conducting sheet. This is known as frequency selective surface (FSS) technology.
Application of thin films and FSS elements permit the signature of a surface to be changed in a deterministic manner. In the proposed work, FSS technology is used in conjunction with the semiconductor device technology for obtaining the desired change in the spectral signature.
J.2 Proposed design concept
The optical properties mainly the relative permittivity (εr) of silicon is strongly affected by injection of charge carriers into an undoped sample or by the removal of free charge carriers from a doped sample. This effect was also observed with the external electric field applied to an undoped silicon sample (the Franz-Keldysh effect), however the desired tunability is very limited and much restricted in case of silicon. The current proposed work utilizes the basic semiconductor device technology concepts where the
194 variation in the free charge carriers is achieved by using a schottky diode structure and external bias. The top FSS structure provides selection of particular wavelength. The change in the thickness of the depletion region (different permittivity) formed between the metal-semiconductor region with the applied bias provides the necessary modulation of the wavelength.
The proposed structure as shown in Figure J.1 is a schottky diode structure consisting of a metal semiconductor junction. This schottky structure is designed so as to get the maximum change in the depletion layer thickness (resonant cavity) with the applied bias. Therefore the design parameters included the barrier height between the contact metal and the semiconductor (p-type silicon with specific dopant concentration), dopant concentration, absorptivity and thickness of the epilayer. Using the basic semiconductor device physics concepts one can obtain the thickness of the depletion with which is given by equation F.1. The calculations are based on the assumptions of an abrupt junction.
1 2 ⎡ ⎡ 2 ⎤⎤ ⎡ 2 2 ⎛ 2 ⎞⎤ ⎢2⋅ε0 ⎢⎛ Eg kT⋅ ⎛ Nv ⎞ ⎞ ⎢ e ⋅λ ni Na ⎥ ⎡⎛⎛ 1 ⎞⎞⎤⎥⎥ Wp() Na := ⋅ ⎜ ⋅− ln⎜ ⎟ +φVa −χm + ⎟⋅ nu + ⋅⎜ + ⎟ ⋅⎢⎜⎜ ⎟⎟⎥ ⎢ e ⎢⎝ e e ⎝ Na ⎠ ⎠ ⎢ 2 2 ⎝ Na⋅ mce mch ⎠⎥ ⎣⎝⎝ Na ⎠⎠⎦⎥⎥ ⎣ ⎣ ⎣ 8⋅π ⋅c ⋅εnu⋅ 0 ⎦ ⎦⎦
…. J.1
Where Wp(Na) is the depletion width on the p-type silicon epilayer, k is the Boltzmann constant, T is the room temperature in Kelvin, Eg is the energy gap of silicon, Nv is the density of states for holes, e is the electronic charge, φm is the work function of the metal,
Va is the applied bias and χ electron affinity in vacuum.
195 A periodic moment method (PMM) was used to simulate the FSS spectral response using the specified device design parameters. Table J.1 shows the parameters used for this simulation.
FSS Slot Antenna Structure of Gold
200nm WF =150nm
Schottky Contact Ti/Al/Ag Layer WS= 100nm V = Va Va - Vi WD= 10nm to 250nm Depletion Layer (ε1) E = V = V ii WD Va Lightly doped p-type Silicon (ε2) WE-WD = 500-(10nm to 250nm)
V = 0 WC= 500nm
Figure J.1. Tunable FSS structure in Si using laser doping technique.
WE = Epilayer thickness, WD = Depletion layer thickness, Va = applied voltage = bias voltage (Vb), Vi = Voltage at the interface of the epilayer and depletion region and
()− WW DE ≈ VV bi , Vb = Bias voltage across the epilayer. WE
196 Table J.1. Parameter used for simulation of tunable FSS using PMM
W (nm) E 2000 500 300
W (nm) D 10 200 10 200 250 10 250
W – W (nm) E D 1990 1800 490 300 250 290 50
V /V i b 0.995 0.9 0.98 0.6 0.5 0.967 0.167
1 - V /V i b 0.005 0.1 0.02 0.4 0.5 0.033 0.833
E/ Vb = (1 - Vi -4 -4 -3 -3 -3 -4 -3 /Vb)/WD 5×10 5×10 2×10 2×10 2×10 1×10 3×10
J.3 Quantum mechanical interpretation of FSS performance
12 11.73 11.4 11
10.38
10.1 9.81 9.81 10 A1 B1
9.38 9.34 C1 A2 8.9 9
Wavelength (microns) Wavelength B2 8.4 C2 8 2000 500 500 300 500 (50nm Mn layer) Silicon Epilayer Thickness (nm)
WD (10nm) WD (200nm) WD (250nm)
Figure J.2. Results of the PMM based on the depletion region width and other device parameters.
197 As a first approximation, the electric field across the epilayer is EW = Va/WE and
Va the electric field per unit thickness of the epilayer is E′W = . So the electric × WW EE field across the depletion region would be
× WV Da = ′WD WEE D =× × WW EE
J.2
Eq. (J.2) indicates that the electric field across the depletion region will decrease as the depletion width decreases. This is expected because ED = (Va – Vi)/WD and Va tends to
Vi as WD decreases. Actually Va tends to Vi faster than ED tending to zero. So ED would be zero as WD tends to zero.
Case 1 (Same WE but different WD): Consider points A1 and A2 in the graph (Figure J.2) for which the epilayer thickness is 500 nm. The depletion layer thickness for point A1 is less than that for point A2. So the electric field across the depletion region will be less for point A1 than that for point A2. Therefore the electrons will be excited to a lower energy level in the case of A1 than in the case of A2, and consequently the energy emitted in the case of A1 will be less than in the case of A2. So the wavelength will be higher in the case of A1 than in the case of A2.
Case 2 (Same WE but different WD): Consider points A2 and B2 for which the epilayer thickness is the same. The depletion layer thickness for point A2 is less than that for point B2. So the electric field across the depletion region will be less for point A2 than that for point B2. Therefore the electrons will be excited to a lower energy level in the case of A2 than in the case of B2, and consequently the energy emitted in the case of A2
198 will be less than in the case of B2. So the wavelength will be higher in the case of A2 than in the case of B2.
Case 3 (Same WD but different WE): Consider points B2 and C2 for which the depletion layer thickness is the same. The epilayer thickness for point B2 is higher than that for point C2. So the electric field across the depletion region will be less for point B2 than that for point C2 due to Eq. (1). Therefore the electrons will be excited to a lower energy level in the case of B2 than in the case of C2, and consequently the energy emitted in the case of B2 will be less than in the case of C2. So the wavelength will be higher in the case of B2 than in the case of C2.
λ1, λ2, λ3, λ4, λ5, ….λn
Evanescent Waves
Schottky Contact Ti/Al/Ag Layer Transmitted Depletion Layer (ε ) Resonance 1 Wavelengths
Drude Zener Theory Lightly doped p-type Silicon (ε2) FSS Array Structure (RLC Circuit) Absorption of Transmitted and Reflection Reflected wavelengths and Evanescent waves Quantum Mechanical Effect
Applied External Output Wavelength Bias
Figure J.3. Mechanisms and concepts for obtaining a tunable FSS structure.
199 Table J.2. Laser doping parameter used for fabrication of Schottky diode for tunable FSS
Dopant Source Sample Substrate Pulse Beam Laser Scanning No. of and Repetitio Spot Fluence speed Passes Carrier n Rate Diameter (J/cm2) (mm/sec) Gas (kHz) (μm) p-type Si (100), ρ-1- TMA, A 10 Ω-cm, 10 112 5.58 0.8 1 Argon at 410 μm 30 psi Thick p-type Si (100), ρ-1- TMA, B 10 Ω-cm, 10 112 6.09 0.8 1 Argon at 410 μm 30 psi Thick p-type Si (100), ρ-1- TMA, C 10 Ω-cm, 10 112 7.105 0.8 1 Argon at 410 μm 30 psi Thick p-type Si (100), ρ-1- TMA, D 10 Ω-cm, 10 112 8.12 0.8 1 Argon at 410 μm 30 psi Thick
Figure J.3. shows the summary of this work and concepts for achieving a tunable FSS structure and Table J.2 shows the laser doping parameters used for fabrication of
Schottky device.
200 LIST OF REFERENCES
Achtziger N. and Witthuhn W. (1997), “Deep levels of chromium in 4H-SiC“, Mater.
Sci. Eng. B, Vol. 46, Issues 1-3, pp. 333-335.
Achtziger N, Witthuhn W. (1998), “Band-gap states of Ti, V, and Cr in 4H-SiC:
Identification and characterization by elemental transmutation of radioactive
isotopes”, Phys. Rev. B, Vol.57, pp 12181-12196.
Aristov V. Y., Douillard L., Fauchoux O. and Soukiassian P. (1997), “Temperature-
Induced Semiconducting c(4×2) ⇔ Metallic (2×1) Reversible Phase Transition in
the β-SiC (100) Surface”, Physical review letters, Vol. 79, No. 19, pp 3700-3703.
Bäuerle D. (2002), Laser processing and chemistry. 3rd. ed., Berlin: Springer,. p. 103 and
117.
Berch A. A. and Dean P. J. (1976), “Light-emitting diodes”, Clarendon Press-Oxford
Publication, UK.
Bet S. M, Quick N. and Kar A. (2008), “Laser Doping of Chromium as a Double
Acceptor in Silicon Carbide with Minimum Crystalline Damage and Maximum
Ionized Dopants”, Acta Materilia, Vol. 56, pp. 1857-1867.
Bet S. M, Quick N. and Kar A. (2008), “Laser-Doping of Chromium in 6H-SiC for White
Light Emitting Diodes”, Journal of Laser Applications, Vol. 20, 1, pp. 1-7.
Bet S. M, Quick N. and Kar A. (2007), “Effect of Laser Field and Thermal Stress on
Diffusion in Laser Doping of SiC”, Acta Materilia, Vol. 55, pp. 6816-6824.
Bet S. M, Quick N. and Kar A. (2007), “Laser-Doping of Silicon Carbide for p-n
Junction and LED Fabrication”, Physica Status Solidi (a), Vol. 204, Issue 4, pp.
1147-1157.
201 Bet S, Quick N and Kar A. (2007), “Laser Doping of Selenium and Chromium in p-type
4H-SiC”, International Conference on Silicon Carbide and Related Materials
2007, October. 14-19, Otsu, Japan.
Bet S, Quick N and Kar A. (2007), “Silicon Carbide White Light LEDs for Solid-State
Lighting”, SPIE Photonic West 2007, Light-Emitting Diodes: Research,
Manufacturing, and Applications XI, Klaus P. Streubel, Heonsu Jeon, Editors,
Proceedings of SPIE -Volume 6486, 648613-1-8, January 20-25, San Jose,
California.
Bet S, Quick N and Kar A. (2006), “Laser Fabrication Of Silicon Carbide Light Emitting
Diodes”, 25th International Congress on Applications of Lasers & Electro-Optics
2006, October 30 – November 2, Scottsdale, Arizona.
Bet S, Quick N and Kar A. (2006), “Laser-Patterned SiC Blue-Green LED”, Materials
Research Society Symposium 2006, Silicon Carbide 2006 - Materials, Processing
and Devices, April 17-21, San Franscisco, California.
Bet S. and Kar A. (2006), “Thin film deposition on plastic substrates using silicon
nanoparticles and laser nanoforming”, Materials Science and Engineering B,
Volume 130, pp 228-236.
Bet S. and Kar A. (2006), “Laser Deposition of Silicon Films using Nanoparticle
Precursor”, Journal of Electronic Materials, Volume 35, Issue 5, pp 993-1004,
2006.
Bet S. and Kar A. (2006), ““Deposition of Crystalline Layers on Polymer Substrates
Using Nanoparticles and Laser Nanoforming”, patent pending 2006.
202 Brander R. W. (1972), “A review of the merits of direct and indirect gap semiconductors
for electroluminescence devices”, Review of Physics in Technology, Issue 3, pp
145-194.
Brander R. W. and Sutton R. P. (1969), “Solution grown SiC p-n junctions”, Journal of
Physics D: Applied Physics, Vol. 2, pp 309-318.
Casey H. C., Jr. and Panish M. B. (1978), “Quantum electronics; Principles and
applications”, Academic Press, NY, Ch. 2.
Castagna M. E, Coffa S., Monaco M., Muscara A., Caristia L., Lorenti S., and Messina
A. (2003), “High Efficiency Light Emission Devices in Silicon”, Mat. Res. Soc.
Symp. Proc. Vol. 770, I2.1.1.
CIE (1932), Commission internationale de l'Eclairage proceedings, Cambridge
University Press, Cambridge, UK.
Cohen M. M. (1975), “Tunneling Electroluminescent Diode with Voltage Variable
Wavelength Output“, US Patent # 371,344.
Craford M. G. (1992), “LEDs Challenge the Incandescents”, IEEE Circuits and Devices
Magazine, September, pp 24-29.
Denbaars S., Tarsa E. J., Mack M., Keller B. and Thibeault B. (2006), US Patent
7,084,436.
Dmitriev V. A., Morozenko Y. V., Popov I. V., Surorov A. V., Syrkin A. L., Chelnokov
V. E. (1986), Soviet Technical Physics Letters, Vol. 12, Issue 5, pp 221-222.
Dmitriev V. A., Kogan L. M., Morozenko Y. V., Tsarenkov B. V., Chelnokov V. E. and
Cherenkov A. D. (1989),” Violet-light-emitting SiC-4H Diodes”, Soviet Physical
Semiconductors, Vol. 23, Issue 1, pp 23-25.
203 Edmond J. A. (1995), “Blue light-emitting diodes with high external quantum
efficiency”, Patent # 5416342, Cree Research, Inc., May 16.
Edmond J., Kong H., Suvorov A., Waltz D. and Carter C., Jr. (1997), “6H-Silicon carbide
light emitting diodes and UV photodiodes”, Phys. Stat. Sol. (a), Vol. 162, pp.
481-491.
Feng Z. C. and Zhao J. H. (2004),” Silicon Carbide: Materials, Processing, and Devices”,
Optoelectronic properties of semiconductors and superlattices, Vol. 20, Taylor
and Francis Books, Inc.
Futagi T., Matsumoto T., Katsuno M., Ohta Y., Mimura H and Kitamura K. (1993), “
Visible light emission from a pn junction of porous silicon and microcrystalline
silicon carbide”, Appl. Phys. Lett. Vol. 63, Issue 9, pp. 1209-1210.
Harris G. L. (1995),” Properties of Silicon Carbide”, Emis Data Review Series No. 13,
Inspec Publication.
Hecht E. (2006), “Optics”, Fourth edition. Pearson education Inc. UK.
Hemstreet A. (1977), “Electronic states of simple-transition-metal impurities in silicon”,
Phys. Rev. B, Vol.15, Issue 2, pp. 834-839.
Humphreys C. J. (2008), “Harnessing Materials for Energy: Solid State Lighting”, MRS
Bulletin, Vol. 33, 459-470.
Hoffmann L., Ziegler G., Theis D. and Weyrich C. (1982),” Silicon carbide blue light
emitting diodes with improved external quantum efficiency”, Journal of Applied
Physics, Vol. 53, Issue 10, pp 6962-6967.
Ikeda M., Hayakawa T., Yamagiwa S., Matsunami H., and Tanaka T. (1979),”Fabrication
of 6H-SiC light emitting diodes by rotation dipping technique:
204 Electroluminescence mechanisms”, Journal of Applied Physics, Vol. 50, Issue 12,
pp 8215-8225.
Justo J. F., Machado W. V. M. and Assali L.V.C. (2006), “Behavior of 3d-transition
metals in different SiC polytypes”, Physica B, Vol. 376-377, pp. 378-381.
Kamiyama S., Maeda T., Nakamura Y., Iwaya M., Amano H. and Akasaki I. (2006),
“Extremely high quantum efficiency of donor-acceptor-pair emission in N and B
doped 6H-SiC”, Journal of applied physics, Vol.99.
Kempshall B. W., Giannuzzi L. A., Prenitzer B. I., Stevie F. A. and Da S. X. (2002),
“Comparative evaluation of protective coatings and focused ion beam chemical
vapor deposition processes”, J. Vac. Sci. Technol. B, 20 (1), 286-290.
Kholuyanov G. F. (1969), “The roles of boron, nitrogen and gallium in the
electroluminescence of silicon carbide”, Soviet Physics-Solid State, Vol. 7, Issue
11, pp 2620-2624.
Krasnov Y. S., Kmita T. G., Ryzhikov I. V., Pavlichenko V. I., Sergeev O. T., and
Suleimanov Y. M.(1969), “ Luminescence of nitrogen-aluminum complexes in α-
SiC (6H)”, Soviet Physics-Solid State, Vol. 7, Issue 4, pp 905-910.
Lang D. V. (1974), “Deep-level transient spectroscopy: a new method to characterize
traps in semiconductors”, J. App. Phy., Vol. 45, pp. 3023-3032.
Lebedev A. A. (1999), “Deep level centers in silicon carbide: A review”,
Semiconductors, Vol. 33, Issue 2, pp.107-130.
Lee K. M., Dang L. S., Watkins G. D., and Choyke W. J. (1985), Physical review B, Vol.
32, pp. 2273.
205 Levinshtein M. E., Rumyantsev S. L. and Shur M. S. (2001),” Properties of advanced
semiconductor materials”, Wiley-Interscience Publication.
Macfarlane P. J. and Stahlbush R. E. (2001), “Characterization of light emission from 4H
and 6H SiC MOSFETs”, Mat. Res. Soc. Symp. Proc. Vol. 640, H4.9.1.
Mimura H., Matsumoto T., Kanemitsu Y. (1994), “Blue electroluminescence from porous
silicon carbide”, Appl. Phys. Lett. Vol. 65, Issue 26, pp. 3350-3352.
Morkoç H. and Mohammad S. N. (1997) , “Light Emitting Diodes” in Wiley
Encyclopedia of Electrical Engineering and Electronics Engineering, Ed. J.
Webster, John Wiley and Sons, Inc. Publishers.
Nakashima S., Kato T., Nishizawa S., Mitani T., Okumura H. and Yamamotob T., J.
(2006),” Deep Ultraviolet Raman Microspectroscopic Characterization of
Polishing-Induced Surface Damage in SiC Crystals”, Electrochem. Soc., 153, 4,
G319-G323.
Neudeck P. G. (2000), “SiC Technology”, Boca Raton (FL): CRC Press. p. 6.1.
Okamoto H., and Massalski T. B. (1985), “The Au−Pd (Gold-Palladium) system”, Bull.
Alloy Phase Diagrams; 6 (3), 229-235.
Okojie R S., Xhang M., Pirouz P., Tumakha S., Jessen G. and Brillson L. J. (2002), “4 H
- to 3 C-SiC polytypic transformation during oxidation”, Materials Science
Forum, 389-393, 451-454 .
Oswald S. and Wirth H. (1999), “Core-level shifts at B- and Al-doped 6H-SiC studied by
XPS”, Surf. Int. Anal. 27, 136-141.
Pan J. (2006), “ Use of deep-levels transitions in semiconductor devices”, US Patent #
7049641.
206 Pankove J. I. (1971), “Optical Processes in Semiconductors”, Prentice Hall Series in
Solid State Physical Electronics, 1971.
Panknin D, Wirth H, Mücklich A, Skorupa W. (2001), “Electrical and microstructural
properties of highly boron-implantation doped 6H-SiC”, J App Phys, Vol.89,
Issue 6, pp.3162-3167.
Pearton S. J. (2000), “Processing of Wide Band Gap Semiconductors”, 1 st. Edition
William Andrew publishing.
Pensel T. G. and Choyke A. D. (1993), “Electrical and optical characterization of silicon
carbide”, Physica B: Condensed Matter, Vol. 185, pp.264-183.
Potter R. M., Blank J. M. and Addamiano A. (1969),”Silicon Carbide Light Emitting
Diodes”, Journal of Applied Physics, Vol. 40, Issue 5, pp 2253-2257.
Phillips J. M., Coltrin M. E, Crawford M. H, Fischer A. J, Krames M. R, Mueller-Mach
R, Mueller G. O, Ohno Y, Rohwer L. E. S., Simmons J. A., and Tsao J. Y. (2007),
“Research challenges to ultra-efficient inorganic solid-state lighting”, Laser &
Photon. Rev. 1, No. 4, 307–333.
Press Release. (2006), “Cree Delivers the First 160-Lumen White Power LED”,
www.cree.com/press release, October 9.
Quick N., Bet S. and Kar A. (2007), “Laser Doping Fabrication of Energy Conversion
Devices”, Materials Science and Technology Conference 2007, September 16-20,
Detroit, Michigan.
Rahman M. M. and Furukawa S. (1992),” Silicon Carbide Turns on its Power”, IEEE
Circuits and Devices Magazine, pp 22-26.
207 Ramachandran V. and Feenstra R. M. (1999), “Scanning Tunneling Spectroscopy of
Mott-Hubbard States on the 6H-SiC (0001) √3×√3 Surface”, Physical Review
Letters, Vol. 82, No. 5, pp. 1000-1003.
Reshanov S.A., Schneider K., Helbig R., Pensl G., Nagasawa H., and Schoner A. (2004),
“Midgap Defects in 4H-, 6H- and 3C-SiC Detected by Deep Level Optical
Spectroscopy”, Mat. Sci. Forum., Vols. 457-460, pp. 513-516.
Reshanov S. A., Pensl G. (2006),“ Sulfur and Selenium Double Donors in SiC“,
University of Erlangen-Nürnberg, ECSCRM, MoP 34.
Robinson, S. J. and Schmidt, J. T. (1984), “Fluorescent Penetrant Sensitivity and
Removability - What the Eye Can See, a Fluorometer Can Measure”, Materials
Evaluation, Vol. 42, No. 8, pp. 1029-1034.
Salama I.A., Quick N.R. and Kar A. (2002), “Laser Doping of Silicon Carbide
Substrates”, Journal of Electronic Materials, Vol. 31, pp. 200-208.
Salama I.A., Quick N.R. and Kar A. (2004), Laser Direct Writing and Doping in
Polycrystalline Diamond and Single Crystal Silicon Carbide, Journal of Laser
Applications.
Salama I.A., Quick N.R. and Kar A. (2003), Microstructural Effects on Electrical
Resistance in Laser-Treated Silicon Carbide, Journal of Materials Science, .
Salama I.A., Quick N.R. and Kar A. (2002), Laser Surface Modification of Electronic
Properties in Wide Bandgap Materials, in Surface Engineering: Science &
Technology II, Edited by Ashok Kumar, Yip-Wah Chung, John J. Moore, Gary L.
Doll, Kyoshi Yatsui and D.S. Misra. Published by TMS (The Minerals, Metals &
Materials Society), pp 113-124 (Invited Paper).
208 Salama I. A. (2003), Ph.D. Dissertation, University of Central Florida, Spring.
Schubert F. E. (2003), “Light Emitting Diodes”, 1st. Edition, Cambridge University Press,
Cambridge, UK.
Sengupta D. K, Quick N. R. and Kar A. (2001), “Laser Conversion of Electrical
Properties for Silicon Carbide Device Applications”, J. Laser Appl. Vol. 13, pp.
26-31.
Shewmon P. G. (1963), “Diffusion in Solids”, New York (NY): McGraw-Hill
publication; p.30.
Simon L. (1970),” Silicon Carbide”, edited by Frantsevich I. N, Academy of Sciences of
Ukranian SSR, USSR.
Streetman B. G. (1990), “Solid state electronic devices”, Third edition, Prentice Hall
Series in solid state physical electronics, pp 151.
Suzuki A., Farukawa K. and Fujii Y. (1991), “ Silicon carbide light emitting diode and a
method for the same”, Patent # 5243204, Sharp Kabushiki Kaisha, Osaka, Japan,
May 17.
Tairov Y. M, V. F. Tsvetkov. (1988), "Semiconductor Compounds AIV BIV", Handbook
on Electrotechnical Materials Eds. Koritskii, Yu.V., V.V. Pasynkov, B.M. Tareev,
Vol.3, Sec.19, , Energomashizdat, Leningrad, 446-471. (in Russian)
Takano K, Nitta H, Seto H, Lee CG, Yamada K, Yamazaki Y, Sato H, Takeda S, Toya E,
Iijima Y. (2001), “Volume and dislocation diffusion of iron, chromium and cobalt
in CVD β-SiC”, Sci Tech Adv Mat, Vol. 2, Issue 2, pp.381-388.
Tian Z., Quick N. R. and Kar A. (2006), “Laser-enhanced diffusion of nitrogen and
aluminum dopants in silicon carbide”, Acta Materialia, Vol. 54, pp. 4273-4283.
209 Tian Z., Quick N. R. and Kar A.(2005), “Effects of different laser sources and doping
methods used to dope silicon carbide”, Acta Materialia, Vol. 53, pp. 2835- 2844.
Tian Z., Quick N. R. and Kar A. (2005), “ Characteristic of 6H-silicon carbide pin diode
prototyping by laser doping”, Journal of Electronic Materials, Vol. 34, Issue 4, pp
430-438.
Tian Z. (2006), Ph.D. Dissertation, University of Central Florida, Summer.
Tian Z., Quick N. R., and Kar A.( 2006), “Laser Endotaxy and PIN Diode Fabrication of
Silicon Carbide, Symposium B: Silicon Carbide: Materials, Processing and
Devices, edited by M. A. Capano, M. Dudley, T. Kimoto, A. R. Powell and S.
Wang, Mater. Res. Soc. Symp. Proc, 911, B10-07.
Violin E. E., Kalnin A. A., Pasynkov V. V., Tairov Y. M., Yaskov D. A. (1969), Silicon
Carbide - 1968, Pergamon Press, 231-241.
Vishnevskaya B. I., Dmitriev V. A., Logan L. M., Morozenko Y. V., Chelnokov V. E.
and Cherenkov A. D. (1990),” Green SiC-6H light-emitting diodes”, Soviet
Technical Physics Letters, Vol. 16, Issue 12, pp 908-909.
Vlaskina S.I. (2002), Silicon carbide LED, Semiconductor Physics, Quantum Electronics
and Optoelectronics, Vol. 5, pp. 71-75.
Wai Lek Ng, Lourenco M. A.¸ Gwilliam R. Ledain M., S., Shao G. and Homewood K. P.
(2001), Nature, Vol. 410, Issue 8, pp.
Woodall J. M., Kornegay K. T. and Spencer M. G. (1998), “Incandescent light energy
conversion with reduced infrared emission”, Patent # 5,814,840.
210 Zhang Q. and Sudarshan T. S. (2001), “Forward characteristics of P+PN+ and P+NN+
diodes on 6H-SiC”, Solid state electronics, 45, pp 1559-1563.
Zetterling C-M, Östling M. (2002), In: Zetterling C-M editor. Process technology for
silicon carbide devices. London: INSPEC, Institution of Electrical Engineers. p.4.
211